# Foxora — complete knowledgebase > Foxora is the first L5 autonomous multi-agent system. Describe an outcome in plain words; a crew of AI specialists plans it, builds it, checks its own work, and delivers the finished thing — with permanent shared memory. It runs as a desktop app, in the web, and as a CLI. Built by xBesh Labs, LLC. Free to start, no credit card, no separate AI accounts. > This single file is the entire Foxora platform — the marketing site and the full product documentation — concatenated as Markdown. Hand it to an AI model as the knowledgebase for answering anything about Foxora. # Part 1 — Site --- source: https://foxora.ai/ title: Foxora — your AI workforce --- Eyebrow: The first L5 autonomous multi-agent system. Headline: Build your team. On foxora. A team that never sleeps and never forgets. Just describe the outcome — a crew of AI specialists researches, builds, verifies, and hands you work that's actually finished, not a draft you have to fix. ## Why Foxora — autonomy has levels AI tools help you work; Foxora works for you. Borrowed from the self-driving levels: L1 autocomplete, L2 copilot (you steer), L3 single agent (you watch), L4 crew (a human gates each step), L5 workforce — you give the goal and it delivers the finished outcome. Foxora is built L5-first, with a full audit trail so autonomy never means opacity. ## How it works One sentence in, a finished thing out. A planner turns your goal into a work plan, specialists run the pieces in parallel, a checker challenges everything, and the result comes back done — with its memory updated. ## The crew A specialist for every job (researcher, coder, reasoner, verifier, scout, tester, writer, designer, data, vision, web-search, devops, security, integrator, and more) — and any role you invent. Roles run in parallel, nothing reaches you before the checker signs off, and each specialist routes to the level of AI its work deserves. ## Queue & delegation Stack up tasks like a to-do list. The planner assigns each to the right crew, delegates across agents and tools, and runs it in the cloud — laptop optional. ## Memory (the Den) It remembers like a colleague, not a chatbot. Your preferences, projects, and rules — explained once, kept for good. Facts settle through four layers (raw, episodic, semantic, relational), and every memory carries a receipt: ask "why do you think that?" and it shows you where it learned it. The Den lives on your machine. ## Tell it once Sessions and memory are one shared record across desktop, web, and CLI. Start anywhere, continue anywhere — never paste the same context twice. ## Skills Done-for-you playbooks (launch a product, run a newsletter, sort a backlog, audit a codebase) — plus a Skill Builder to teach Foxora your own way of doing things. A skill is a procedure with tools and scripts attached. ## Your workforce Build departments, crews, and agents like a real company. Or describe what you need and the AI builder drafts the whole team. You stay the boss. ## 1,000+ plugins Connected to the apps you already use — Gmail, GitHub, Slack, Stripe, Linear, Notion, and more, one click each. OAuth tokens are held by Composio (the connector-auth layer trusted by teams like Anthropic and OpenAI), never on Foxora's servers. Revoke anything, any time. ## Built-in IDE A full code studio inside: tabs, files, git (stage, commit, PR), and a live Stage where agents open terminals and browsers. Multiple projects open at once. If you don't code, you never see this screen. ## Voice Talk to your crew like a chief of staff. Ask what happened overnight, order what ships next — it answers out loud, queues the work, and reports back. ## Everywhere The same crew and memory in a window (Desktop), a browser (Web, app.foxora.ai), or a terminal (CLI). One account, everywhere. ## Overnight (cloud agents) Scheduled crews run in the cloud — they don't need your laptop open, and everything they learn lands in the same memory. ## No API keys, ever Your plan includes the intelligence, metered in simple credits like minutes on a phone plan. Every request routes to the right level of AI (Fast, Premium, Max) — no provider secrets on your machine; sign-in is PKCE end to end. ## Your work stays yours The desktop engine runs on your machine; your files never have to leave, and nothing trains on them. Every action is logged and rewindable. Signed builds, public releases, and an install script you can read before you run it. ## FAQ (summary) - Do I need to code? No — talk in plain words; builders get a code studio, everyone else never sees it. - What is L5? Full autonomy: you give the destination, it drives; every action logged. - Mistakes? A checker validates before you see it; every change is a reviewable diff; stop/rewind anytime. - AI accounts / API keys? None — the plan includes the intelligence, metered in credits. - Where do OAuth tokens live? In Composio's vault, never on Foxora's servers; revoke anytime. - Where does my work live? On your machine (desktop); cloud work runs in inspectable isolated sessions; PKCE sign-in. --- source: https://foxora.ai/pricing title: Foxora — Pricing --- Simple, credit-based pricing for your AI workforce. One plan covers the whole crew — every model, every surface. No API keys, no per-provider bills, no surprises. Start free, then Starter, Pro, or Max. ## Plans ### Free — Free Kick the tires — no card required. Monthly credits: 200. ### Starter — $29/mo, or $24/mo billed annually ($290/yr) For individuals and small projects. Monthly credits: 2,000. ### Pro — $59/mo, or $49/mo billed annually ($590/yr) (recommended) For professionals and growing teams. Monthly credits: 5,000. ### Max — $199/mo, or $166/mo billed annually ($1,990/yr) Maximum power for heavy users and agencies. Monthly credits: 20,000. Annual billing includes 2 months free. Credits meter everything like minutes on a phone plan — one grant per month, spent across the whole crew. Full side-by-side feature comparison is on the pricing page. --- source: https://foxora.ai/download title: Foxora — Download --- Get Foxora on your desktop, in your browser, or in your terminal. ## Desktop app macOS (Apple Silicon & Intel, signed and auto-updating) and Windows 10/11. Linux desktop is coming soon. Builds are signed and update themselves from the public releases repo. ## Web app Skip the install entirely — open app.foxora.ai in any browser. Same crew, same memory, same account. ## macOS installer / updater One command installs or replaces Foxora Desktop and then provisions Foxora Core: `curl -fsSL https://foxora.ai/install.sh | sh`. It preserves your account, projects, sessions, and Memory Den. ## CLI The CLI has its own installer: `curl -fsSL https://foxora.ai/install-cli.sh | sh` on macOS/Linux, or `irm https://foxora.ai/install.ps1 | iex` on Windows. Three steps to start: install (or open the web app) → sign in at app.foxora.ai → describe a goal. No API keys, no separate AI accounts. --- source: https://foxora.ai/cli title: Foxora — CLI --- A crew of agents in your terminal. Install: `curl -fsSL https://foxora.ai/install-cli.sh | sh` (macOS/Linux) or `irm https://foxora.ai/install.ps1 | iex` (Windows). The CLI installs the same engine as the desktop app; the runtime provisions on first run. ## Command reference - `foxora` — open the interactive terminal UI - `foxora setup` — provision the engine, Memory Den, and Bun runtime - `foxora update` — update the CLI and runtime - `foxora login` / `foxora logout` — sign in or out - `foxora run [crew] -n ""` — run one prompt and print the reply - `foxora usage` — show weekly and monthly credit usage - `foxora doctor` — diagnose the engine, runtime, and authentication - `foxora models` — list available tiers and models - `foxora whoami` — show the signed-in account - `foxora sessions` — list sessions - `foxora crews` — list crews - `foxora agents` — list agents - `foxora --help` — show command usage, options, and environment variables ## Built for pipelines Headless runs stream plain text to stdout and return meaningful exit codes. They work with shell redirection and logging in ssh, Docker, CI, cron, Makefiles, and git hooks. One account, one memory — the same crew as desktop and web. --- source: https://foxora.ai/status title: Foxora — Status --- Live status of every Foxora service. Health is checked directly from our own servers (no third-party monitoring): the web app (app.foxora.ai), AI gateway (api.foxora.ai), cloud engine (engine.foxora.ai), auth & database (Supabase), docs, and the website. Each service shows its current state (operational / degraded / down) and response time, refreshed every minute. Contact: status@foxora.ai. --- source: https://foxora.ai/about title: Foxora — About --- Foxora is the operating system for AI agents, a product of xBesh Labs, LLC — a Delaware limited liability company registered at 8 The Green, Dover, Delaware, United States 19901. Mission: make AI agents practical, observable, and production-ready — real systems that run your business, not demos. State the goal; ship the thing. Contact: hello@foxora.ai (partnerships, media, general inquiries). --- source: https://foxora.ai/brand title: Foxora — Brand --- The Foxora brand kit. - Logo: the app-icon tile (dark rounded square holding the two-tone fox) plus the "foxora" wordmark, with SVG downloads. Keep the icon and wordmark together or use the icon alone. - Color: espresso #171412 (background), cream #f5f2ec (ink), and pastel accents — peach, rose, lavender, sage. - Typography: Cal Sans (display), Inter (body), JetBrains Mono (code). - Do: give the mark clear space; use the espresso or cream background; write "Foxora" with a capital F. - Don't: recolor, stretch, rotate, or add effects; place it on a busy background; misspell it ("foXora", "Foxura"). Brand questions: press@foxora.ai. --- source: https://foxora.ai/privacy title: Foxora — Privacy Policy --- We take your privacy seriously. This policy explains exactly what data Foxora collects, why we collect it, how we protect it, and the choices you have. ## 1. Overview xBesh Labs, LLC ("xBesh Labs", "we", "us", or "our") is the legal entity behind Foxora AI. We operate the website at foxora.ai and the Foxora AI platform, including its API, agent runtime, dashboard, and all related services (collectively, the "Service"). xBesh Labs, LLC is incorporated in Delaware, United States. Our registered address is 8 The Green, Dover, Delaware, United States 19901. This Privacy Policy describes how we collect, process, store, share, and protect personal information when you use the Service, and explains the choices available to you regarding your personal data. Plain language summary: We only collect what we need to run the product. We do not sell your data. We give you full control to access, update, or delete your information at any time. ## 2. Data we collect Account & identity — name, email, and password hash when you register; profile settings and preferences; billing name, address, and payment metadata (card last four, expiry — full card numbers are never stored by us, handled exclusively by Stripe); company name and role if provided. Product usage — API requests, model calls, token counts, latency, and cost records; feature flags; agent configurations you create; support tickets and communications. Technical & device — IP address, browser, OS, and device type; log data including error traces, request IDs, and timestamps; session tokens and authentication events; referral source and UTM parameters. Content you provide — prompts, inputs, and outputs processed through the Foxora API (used solely to deliver the service; not used for model training); files or data you upload for analysis features. ## 3. How we use data We use the information we collect to: - Provision, operate, and maintain the Service and its features - Process billing, manage subscriptions, and prevent fraud - Enforce plan limits, rate limits, and acceptable-use policies - Send transactional emails (confirmation, invoices, password reset, usage alerts) - Provide customer support and resolve technical issues - Monitor and improve reliability, performance, and security - Generate aggregated, anonymized analytics about service usage - Comply with legal obligations and respond to lawful requests We do not use your prompts, agent outputs, or API request content to train any machine learning model — ours or any third party’s. ## 4. Data sharing We do not sell, rent, or trade your personal information. We may share data with third parties only in these limited circumstances: Service providers — vendors who help us operate the Service, including Supabase (database & auth), Stripe (payments), Postmark (email), Sentry (error monitoring), Upstash (caching), and Railway (infrastructure), under data processing agreements. AI model providers — when you make API requests, your inputs are routed through our gateway to third-party model providers (e.g. Google, OpenAI) to generate responses. We route through our gateway so providers never receive your Foxora credentials or account identity. Legal requirements — we may disclose information if required by law, court order, or to protect the rights, property, or safety of xBesh Labs, LLC, our users, or the public. Business transfers — in a merger, acquisition, or sale of assets, user data may be transferred. We will notify users via email and a prominent notice before any such transfer. ## 5. Retention & security We retain personal data for as long as your account is active and as necessary to provide the Service, meet legal obligations, resolve disputes, and enforce agreements. - Account data — retained while active and for up to 90 days after deletion, then permanently purged - API usage logs — up to 13 months for billing reconciliation and abuse detection - Payment records — 7 years, as required by financial regulations - Support communications — 3 years Security protections include TLS/HTTPS for data in transit, AES-256 encryption for sensitive values at rest, row-level security and role-based database access, multi-factor authentication for infrastructure access, and regular security reviews and dependency audits. ## 6. Cookies & tracking We use essential cookies and browser storage to operate the Service — session and authentication state (Supabase Auth), dashboard theme preference (localStorage), affiliate tracking cookies (30-day window), and CSRF protection tokens. We do not run third-party advertising networks or cross-site behavioral tracking. You can manage cookies in your browser settings, though disabling essential cookies will prevent login. ## 7. Your rights Depending on your jurisdiction, you may have the following rights: - Access — request a copy of the personal data we hold about you - Correction — update inaccurate profile information via dashboard settings - Deletion — request permanent deletion of your account and data - Portability — export your usage data and account information - Restriction — object to certain processing activities - Opt-out of marketing — unsubscribe from non-transactional emails at any time To exercise these rights, contact us at privacy@foxora.ai or use your dashboard settings. We respond to all requests within 30 days. ## 8. Children The Service is not directed to individuals under 16. We do not knowingly collect personal information from children. If we learn that a child has provided us with personal data, we will delete it promptly. If you believe a child has provided us with data, contact privacy@foxora.ai. ## 9. Policy changes We may update this Privacy Policy from time to time. When we make material changes, we will notify you by email and update the "Last updated" date. Continued use of the Service after notification constitutes acceptance of the updated policy. ## 10. Contact us For privacy-related questions, data requests, or to report a concern, contact us at privacy@foxora.ai. --- source: https://foxora.ai/terms title: Foxora — Terms of Service --- By accessing or using Foxora, you agree to be bound by these terms. Please read them carefully before using the Service. ## 1. Agreement to terms These Terms of Service ("Terms") constitute a legally binding agreement between you ("User", "you") and xBesh Labs, LLC, a Delaware limited liability company ("xBesh Labs", "Company", "we", "us"), the legal entity operating the Foxora AI platform, API, dashboard, and all associated services (the "Service"). xBesh Labs, LLC is registered at 8 The Green, Dover, Delaware, United States 19901. By creating an account, accessing the API, or otherwise using the Service, you acknowledge that you have read, understood, and agree to these Terms and our Privacy Policy. If you are using the Service on behalf of an organization, you represent that you have authority to bind that organization. If you do not agree, do not create an account or use the Service. ## 2. The Service Foxora provides an AI agent platform including a web dashboard, REST API gateway, agent orchestration runtime, and developer tools. The Service allows users to send requests to AI language models through the Foxora API gateway; configure, deploy, and monitor agents; access usage analytics, billing, and team administration; and integrate Foxora capabilities into third-party applications via API keys. We reserve the right to modify, suspend, or discontinue any aspect of the Service at any time, with reasonable notice where possible. We are not liable for any modification, suspension, or discontinuation of the Service. ## 3. Accounts & registration To access the Service, you must register for an account and provide accurate, complete, and current information. You are responsible for maintaining the confidentiality of your credentials (including API keys), all activity under your account, promptly notifying us of any unauthorized access, ensuring only authorized individuals access your account, and keeping your account information accurate. You must be at least 16 years old to create an account. Business accounts must be operated by individuals authorized to enter contracts on behalf of their organization. ## 4. Acceptable use You agree to use the Service only for lawful purposes and in accordance with these Terms. You must not: - Generate, distribute, or facilitate illegal, harmful, or abusive content - Violate any applicable local, national, or international law or regulation - Attempt to circumvent rate limits, plan restrictions, or security controls - Reverse engineer, decompile, or extract source code from the Service - Scrape, crawl, or extract data from the Service beyond normal API usage - Resell, sublicense, or white-label the Service without a written agreement - Impersonate any person or entity or misrepresent your affiliation - Transmit malware, viruses, or disruptive code through the Service - Impose unreasonable load on the infrastructure - Generate spam, phishing content, or deceptive material We reserve the right to investigate and take appropriate action, including suspending or terminating accounts that violate these policies. ## 5. Billing & plans Paid plans are billed on a recurring basis (monthly or annual) via Stripe. Fees are non-refundable except as required by applicable law or at our sole discretion. Subscriptions automatically renew at the end of each billing period unless you cancel before the renewal date; cancellation takes effect at the end of the current paid period. Each plan includes a monthly credit grant, feature and crew access, and structural safety limits. Billable product usage draws from the shared credit balance; new billable work may pause when that balance is exhausted. We may change pricing with at least 30 days’ advance notice, taking effect at the start of your next billing cycle. We do not offer refunds for unused portions of a subscription period or unused or partially used credit grants; exceptions may be made at our discretion for billing errors. ## 6. API usage - API keys must be kept confidential and not committed to version control or exposed in client-side code - You are responsible for all API usage billed to your account, whether authorized or not - Rate limits apply per plan tier and are enforced by the gateway - We may throttle or block API access that appears abusive or violates these Terms - You agree not to use the API to build competing AI gateway or proxy services If you suspect unauthorized use of your API key, revoke it immediately from your dashboard and contact security@foxora.ai. ## 7. Intellectual property The Service, including its software, design, trademarks, and documentation, is owned by xBesh Labs, LLC and protected by intellectual property laws. The brand name Foxora AI and associated marks are trademarks of xBesh Labs, LLC. You retain ownership of any content, data, or configurations you create through the Service. By using the Service, you grant xBesh Labs, LLC a limited license to process and transmit your content solely to provide the Service to you. Outputs generated by AI models through the Foxora API are subject to the terms of the underlying model providers. xBesh Labs, LLC makes no ownership claims over AI-generated outputs delivered to you. ## 8. Warranties & disclaimers THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED. To the fullest extent permitted by law, xBesh Labs, LLC disclaims all warranties including fitness for a particular purpose or merchantability; uninterrupted, timely, secure, or error-free operation; accuracy, reliability, or completeness of AI-generated outputs; and that any defects will be corrected. AI language model outputs may be inaccurate, incomplete, or harmful. You are responsible for reviewing and validating any AI-generated content before acting on it. ## 9. Limitation of liability To the maximum extent permitted by applicable law, in no event shall xBesh Labs, LLC be liable for any indirect, incidental, special, consequential, or punitive damages; loss of profits, revenue, data, or business opportunities; damages from unauthorized access to your data; damages from reliance on AI-generated outputs; or service outages and performance degradation. The total aggregate liability of xBesh Labs, LLC shall not exceed the greater of $100 USD or the amount you paid to xBesh Labs, LLC in the twelve months preceding the claim. ## 10. Termination You may delete your account at any time from dashboard settings. We may suspend or terminate accounts that violate these Terms, with or without notice depending on severity, or for non-payment after reasonable notice. Upon termination, your right to use the Service ceases immediately, and we will retain data for the period described in our Privacy Policy. Sections that by their nature should survive termination (intellectual property, warranty disclaimers, limitation of liability, and governing law) will survive. ## 11. Governing law & disputes These Terms are governed by the laws of the State of Delaware, United States, without regard to conflict of law principles. Any dispute shall first be attempted through good-faith negotiation, and thereafter through binding arbitration on an individual basis in Delaware, waiving any right to a class action. Nothing prevents either party from seeking injunctive or other equitable relief in a court of competent jurisdiction in Delaware where necessary to protect intellectual property or confidential information. ## 12. Contact For legal notices, compliance requests, or contract inquiries, write to legal@foxora.ai. --- source: https://foxora.ai/cookies title: Foxora — Cookie Policy --- This policy explains how and why Foxora uses cookies and similar tracking technologies across our website and platform. ## 1. What are cookies Cookies are small text files placed on your device by websites you visit. They are widely used to make websites work efficiently and to provide reporting information. xBesh Labs, LLC (operating as Foxora AI) uses cookies and similar storage mechanisms (such as localStorage and sessionStorage) for the purposes described below. ## 2. Essential cookies Essential cookies are required for the Service to function. Without them, login, session management, and secure account access are not possible. These cannot be disabled. - FoxoraAuth — Supabase auth session — Session - FoxoraRefresh — session renewal — 1 year - FoxoraTheme — dashboard color theme — Persistent - FoxoraBotProtection — Cloudflare bot protection — 30 min ## 3. Preference storage We store lightweight UI preferences in your browser’s localStorage to improve your experience across visits, including your theme selection. These values are stored locally on your device and are not transmitted to our servers. ## 4. Analytics We collect anonymized usage metrics to understand product performance and reliability. This data is aggregated and cannot be tied to an individual user. We do not use third-party behavioral advertising networks or cross-site tracking. - Page visit counts and navigation patterns (anonymized) - Feature usage rates and error frequencies - Performance timing data to detect degradation ## 5. Affiliate tracking When you arrive via a Foxora affiliate referral link (e.g. foxora.ai?ref=CODE), we store a tracking cookie for 30 days to attribute any resulting conversion to the referring affiliate partner. This cookie records the affiliate referral code, a hashed visitor ID for deduplication, and UTM parameters from the referral link. This data is used only for commission attribution and is not shared with advertising networks. ## 6. Your control You can manage cookie preferences through your browser settings. Most browsers allow you to refuse new cookies, delete existing cookies, or be notified when new cookies are set. Note: disabling essential cookies (Supabase auth tokens) will prevent you from staying logged in. Preference and analytics cookies can be cleared without affecting core functionality. ## 7. Contact For questions about cookies or tracking on our platform, contact us at privacy@foxora.ai or write to xBesh Labs, LLC, 8 The Green, Dover, Delaware, United States 19901. # Part 2 — Documentation --- source: https://docs.foxora.ai/getting-started/overview title: Overview --- # Overview Set up Foxora in the safest order: install the appropriate client, sign in, choose an autonomy default, open a small project, and complete one verified task. ![Getting Started in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. Foxora v5.0.3 has two top-level workspaces: **foxora** for tasks, projects, workflows, and schedules; and **studio** for bots, agents, crews, skills, tools, memory, and plugins. > **Safety:** Download installers only from Foxora or the official release repository. Attach only folders needed for the task. ## Steps 1. Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. 2. Set up Foxora in the safest order: install the appropriate client, sign in. 3. Choose an autonomy default. 4. Open a small project. 5. Complete one verified task. 6. Sign in, confirm version 5.0.3, keep **Ask Permissions** for the first task, and use the smallest project or cloud workspace that contains the required work. 7. **Confirm the result.** Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. ## Choose the right path - Use Desktop for local projects and visual work. - Use the web app when you do not need local files. - Use the CLI for terminal-first or scripted work. ## Confirm it worked - Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/getting-started/introduction-to-foxora title: Introduction to Foxora --- # Introduction to Foxora Understand the relationship between a task, project, agent, crew, Stage, tools, and local Memory Den before configuring advanced features. ![Introduction to Foxora in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. Foxora v5.0.3 has two top-level workspaces: **foxora** for tasks, projects, workflows, and schedules; and **studio** for bots, agents, crews, skills, tools, memory, and plugins. > **Safety:** Download installers only from Foxora or the official release repository. Attach only folders needed for the task. ## Steps 1. Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. 2. Understand the relationship between a task, project, agent, crew, Stage, tools, and local Memory Den before configuring advanced features. 3. **Confirm the result.** Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. ## Choose the right path - Use Desktop for local projects and visual work. - Use the web app when you do not need local files. - Use the CLI for terminal-first or scripted work. ## Confirm it worked - Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/getting-started/installation title: Installation --- # Installation Install Foxora Desktop on Windows or macOS, or use the Foxora CLI and web app on Linux while the Linux desktop build is being prepared. ![Installation in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Before you begin You need a Foxora account and an internet connection for sign-in, updates, model requests, and first-run provisioning. For a comfortable experience, use a computer with at least 8 GB of RAM and approximately 1 GB of free disk space. Download Foxora only from the official Foxora download page or the official `foxora-ai/releases` repository. Do not install a file received through email, chat, or an unofficial mirror. | Platform | Supported option | Package | | --- | --- | --- | | Windows 10/11, x64 | Foxora Desktop and Foxora CLI | `.exe` installer | | macOS 12 or newer | Foxora Desktop and Foxora CLI | Signed universal `.dmg` | | Linux | Foxora CLI and web app; desktop is not currently published | Shell installer | ## Windows Foxora supports 64-bit Windows 11 and Windows 10 version 22H2. The desktop installer is a per-user installation, so a normal installation should not require administrator access. ### Install Foxora Desktop 1. Open the official Foxora download page and select **Windows**. 2. Download the Windows x64 installer it offers — the filename carries the release, for example `Foxora_5.0.4_x64-setup.exe`. 3. Open the installer. Windows may show a SmartScreen warning while Foxora completes code-signing. 4. Follow the installer prompts. Foxora installs the desktop application, Foxora Core, and its browser runtime. 5. If Microsoft Edge WebView2 or the Microsoft Visual C++ x64 runtime is missing, the installer downloads and installs the required component automatically. 6. Launch Foxora from the Start menu or desktop shortcut. > **Windows SmartScreen.** We're in the middle of code-signing for Windows. If you hit a SmartScreen warning, click **More info**, then click **Run anyway**. ### Windows first launch 1. Allow the local Foxora Core service a few moments to start. 2. Select **Sign in**. Foxora opens the secure account flow in your default browser. 3. Complete sign-in and return to the desktop app when the browser prompts you. 4. Finish the system check and review the default autonomy setting. 5. Create a task and send a small request to confirm that the runtime, gateway, and workspace are ready. ### Optional: install the Windows CLI Open PowerShell and run: ```powershell irm https://foxora.ai/install.ps1 | iex ``` Then verify the command and finish setup: ```powershell foxora --version foxora setup foxora login ``` ## macOS Foxora supports macOS 12 Monterey or newer. Every macOS build is notarized. ### Download the current package The download page detects your Mac and offers the right image, so you do not have to choose. Two are published: | Image | Runs on | Why pick it | | --- | --- | --- | | `Foxora__universal.dmg` | Apple Silicon and Intel | The safe choice on any Mac, and the only one that works on Intel | | `Foxora__AppleSilicon.dmg` | Apple Silicon only | Roughly half the download size | If you are scripting an install or are unsure which Mac you are on, take the universal image. ### Install Foxora Desktop 1. Download the appropriate `.dmg` from the official Foxora download page. 2. Open the downloaded disk image. 3. Drag **Foxora** into the **Applications** folder. 4. Eject the Foxora disk image. 5. Open Foxora from **Applications** or Spotlight. 6. Complete sign-in in your browser, then return to Foxora. Foxora’s public macOS build is signed and notarized. On first use, macOS may request access to files, browser control, Accessibility, or Screen Recording when you enable a feature that needs it. Grant only the capabilities you intend to use; you can review them later in **System Settings → Privacy & Security**. ### Optional: install the macOS CLI Run the official shell installer: ```bash curl -fsSL https://foxora.ai/install-cli.sh | sh ``` Open a new terminal, then run: ```bash foxora --version foxora setup foxora login ``` ## Linux The Foxora Desktop package for Linux is not published in the current desktop release. Do not download unofficial AppImage, `.deb`, or `.rpm` packages claiming to be the current Foxora Desktop build. You can use the full web app or install the native Foxora CLI on a glibc-based distribution such as Ubuntu, Debian, Fedora, or Arch Linux. ### Install the Linux CLI 1. Confirm that `curl` and `tar` are installed. 2. Run the official installer: ```bash curl -fsSL https://foxora.ai/install-cli.sh | sh ``` 3. Open a new terminal so the updated `PATH` is loaded. 4. Provision the local runtime and sign in: ```bash foxora --version foxora setup foxora login ``` 5. Start the interactive terminal: ```bash foxora ``` For a one-off task, run: ```bash foxora run "summarize this project" ``` ### Use Foxora without installing Open `https://app.foxora.ai` in a supported browser and sign in with the same Foxora account. The web app is the recommended graphical option on Linux until an official desktop package is published. ## Verify your installation After installation or CLI setup: 1. Confirm that Foxora shows your account and the main Tasks, Projects, and Studio navigation. 2. Create a new task. 3. Attach a small test folder only if the task needs file access. 4. Keep **Ask Permissions** enabled for the first run. 5. Ask Foxora to explain the test folder or answer a simple question. 6. Confirm that the task finishes normally and that any requested permission is shown before the action occurs. ## Common installation problems | Problem | What to check | | --- | --- | | Windows installer does not open | Confirm Windows 10 22H2 or Windows 11, x64. If SmartScreen appears, select **More info**, then **Run anyway**. | | WebView2 installation fails | Check the internet connection, install pending Windows updates, and run the Foxora installer again. | | Browser sign-in does not return to Foxora | Leave Foxora open, finish the browser flow, and allow the `foxora://` link to open the app. | | `foxora` is not found after CLI installation | Open a new terminal and confirm that the installer’s bin directory is on `PATH`. | | Linux desktop download is unavailable | Use the Foxora web app or CLI; the desktop build has not shipped yet. | ## Next steps - Review **System requirements** for platform-specific hardware and software details. - Continue to **First-run setup** to configure permissions, models, and autonomy. - Follow **Quickstart: complete your first task** to verify the complete workflow. - See **Updating Foxora** before changing versions or replacing an existing installation. --- source: https://docs.foxora.ai/getting-started/system-requirements title: System requirements --- # System requirements Choose the Foxora surface that matches your operating system and whether the task needs access to files on your device. ![System requirements in Foxora v5.0.3](/docs/images/v5.0.3/settings/about.webp) ## Supported platforms | Surface | Supported platforms | Current release | | --- | --- | --- | | Foxora Desktop | macOS 12 or newer; Windows 10 22H2 or Windows 11 x64 | 5.0.3 | | Foxora CLI | macOS Apple Silicon or Intel; Linux x64 or ARM64; Windows x64 | 5.0.3 | | Foxora web app | Current Chrome, Edge, Firefox, or Safari | 5.0.3 | The Linux desktop application is not currently published. Linux users should use the web app or native CLI. ## Hardware and storage - **Memory:** 8 GB RAM is the practical minimum; 16 GB or more is recommended for large repositories, browser automation, and multiple running tools. - **Disk:** Keep at least 4 GB free for the desktop application, managed browser, updates, logs, and working data. Projects and generated artifacts require additional space. - **Processor:** Foxora 5.0.3 for macOS is universal and runs on Apple Silicon or Intel. The published Windows desktop and CLI builds require x64. - **Display:** A 1280 × 720 viewport is the minimum comfortable desktop workspace. The web interface also supports narrower mobile viewports for review and lightweight tasks. ## Network requirements Foxora needs HTTPS access to `foxora.ai`, `app.foxora.ai`, `api.foxora.ai`, `engine.foxora.ai`, GitHub Releases, and any providers or plugins you connect. Corporate proxies, VPNs, DNS filters, or outbound firewalls must allow the required WebSocket and server-sent-event connections as well as normal HTTPS traffic. ## Local and web workspace boundaries | Runtime | File access | | --- | --- | | Desktop | Works with folders you explicitly open on the device. | | CLI | Uses the selected local working directory and local Foxora runtime. | | Web | Uses its cloud workspace. It does not automatically expose files or tasks stored only on your computer. | The web runtime creates or resumes an isolated Daytona workspace when cloud execution is needed. New and lifecycle-recreated workspaces use the Foxora v2 image with Node.js, Bun, Git, ripgrep, Chromium, and document-generation support. ## Operating-system permissions Desktop features may request file access, notifications, microphone access, Accessibility, Screen Recording, or browser-control permissions. Foxora requests these only when the corresponding feature needs them. Grant the minimum access required and review it later in macOS **Privacy & Security** or Windows **Privacy & security** settings. ## Verify compatibility 1. Install the appropriate Foxora surface. 2. Open **Settings → System** or run `foxora doctor`. 3. Confirm the runtime, account, engine, browser, and workspace checks that apply to the device. 4. Create a small task and complete one read-only task. 5. If the task needs files, open a test folder and confirm Foxora reports the expected path before allowing edits. Successful installation means the interface opens, authentication completes, the runtime reports healthy, and a small task finishes without an unresolved permission or connection error. --- source: https://docs.foxora.ai/getting-started/first-run-setup title: First-run setup --- # First-run setup Foxora Desktop v5.0.3 uses a short four-step personalization flow. It asks what you do, what you want Foxora to help with, how much autonomy new tasks should inherit, and how the voice agent should address you. The web app does not show this Desktop setup flow. After web sign-in, it opens the workspace directly. ## Before you start - Install Foxora Desktop and complete browser sign-in. - Keep an internet connection available while Foxora loads the account and local runtime. - Treat role and goal answers as personalization, not permissions. They influence suggestions but do not limit the product. ## 1. Choose your role On **What best describes your work?**, choose Developer, Founder, Product / PM, Designer, Marketer, Data / Analyst, Writer, or Something else, then select **Continue**. Foxora uses this answer with your goals to prioritize useful agents, skills, recipes, and starter actions. ## 2. Choose your goals On **What do you want to do with Foxora?**, select every relevant outcome. The current choices cover building software, automation, research, data analysis, content and design, business operations, web tasks, and learning. Select **Continue** after checking the list. These choices do not prevent you from using other workflows later. ## 3. Set the default autonomy On **How hands-on should agents be?**, choose the default for new tasks: | Onboarding choice | Composer mode | Behavior | | --- | --- | --- | | Ask first | Ask Permissions | Requests approval before each file edit and command. | | Auto-approve edits | Accept Edits | Applies routine edits while keeping riskier commands gated. | | Autopilot | Auto mode | Runs a bounded task end to end without routine approval stops. | The onboarding screen initially selects **Ask first**. After onboarding, each new task inherits the saved machine-level default until you change it from the composer. > **Bypass Permissions is not an onboarding option.** It remains available in the composer for trusted, recoverable environments, but should not be the normal first-run choice. ## 4. Configure the voice agent If voice is available for the account, choose: 1. The agent name shown in voice interactions. 2. The voice used for spoken replies. 3. What the agent should call you. 4. A personality tone. Select **Start using Foxora** to save the choices, or **Skip** to keep the existing voice defaults. You can change these later under **Settings → Voice**. ## Verify the workspace ![First-run setup in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) 1. Confirm the **foxora** workspace opens under the expected account. 2. Choose **New Task** and check the autonomy pill below the composer. 3. Open the narrowest project folder required for file work, or stay project-free for questions and planning. 4. Send one small, reversible request. 5. Verify the result in the transcript, Stage, project, or external destination. If the selected role or goals appear wrong, finish the flow and change the task configuration directly. If the autonomy default is too broad, change it from the composer before sending the first request. --- source: https://docs.foxora.ai/getting-started/quickstart-complete-your-first-task title: Quickstart: complete your first task --- # Quickstart: complete your first task Open a small test folder, ask for one clearly measurable change, review the plan and approvals, then verify the changed file or command output. ![Quickstart: complete your first task in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. Foxora v5.0.3 has two top-level workspaces: **foxora** for tasks, projects, workflows, and schedules; and **studio** for bots, agents, crews, skills, tools, memory, and plugins. > **Safety:** Download installers only from Foxora or the official release repository. Attach only folders needed for the task. ## Steps 1. Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. 2. Open a small test folder, ask for one clearly measurable change. 3. Review the plan and approvals. 4. Verify the changed file or command output. 5. **Confirm the result.** Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. ## Choose the right path - Use Desktop for local projects and visual work. - Use the web app when you do not need local files. - Use the CLI for terminal-first or scripted work. ## Confirm it worked - Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/getting-started/updating-foxora title: Updating Foxora --- # Updating Foxora Foxora Desktop and CLI 5.0.3 use signed release artifacts. The web app updates automatically when a new production deployment is loaded. ![Updating Foxora in Foxora v5.0.3](/docs/images/v5.0.3/settings/about.webp) ## Desktop automatic updates Foxora Desktop checks the signed update manifest for a newer compatible version. When an update is available, the sidebar shows **Update available**. 1. Save or finish any active work. 2. Select the update notice, or open **Settings → About → Check for updates**. 3. Review the target version and release notes. 4. Download and install the update. 5. Select **Restart to update** when the package is ready. 6. Return to **Settings → About** and confirm the new version. You can also run **Check for Updates…** from the application menu or Command Center. Foxora verifies the updater signature before installation; do not replace the signed package with an installer from an unofficial mirror. ## Manual desktop update If the in-app check cannot reach the manifest: 1. Confirm the [Foxora status page](https://foxora.ai/status) and your internet connection. 2. Download the current package from the [official download page](https://foxora.ai/download). 3. Quit Foxora after active tasks have finished. 4. Install the new package over the existing application. 5. Reopen Foxora and confirm the version under **Settings → About**. Normal upgrades preserve tasks, settings, authentication, Memory Den data, and project references. Export important local data before troubleshooting with a reset or removal. ## Update the CLI Run: ```bash foxora update foxora --version foxora doctor ``` `foxora update` refreshes the CLI and its managed runtime. The CLI and Desktop can have separate release histories, so verify each surface independently. ## Update the web app The web runtime is deployed centrally. Refresh `https://app.foxora.ai` to load the current production assets. If an old tab remains open across a deployment, finish or save active work, then reload it. ## Verify the update After any update: 1. Confirm the displayed version. 2. Open an existing task and create a new task. 3. Run one small tool or workspace task. 4. If you use plugins, channels, browser control, or scheduled work, test one harmless representative action. 5. Keep the exact version, time, and error if a regression appears. Do not clear local data merely because an update check fails. Diagnose the updater, network, runtime, and account layers separately. --- source: https://docs.foxora.ai/getting-started/uninstalling-and-resetting-foxora title: Uninstalling and resetting Foxora --- # Uninstalling and resetting Foxora Choose between removing the application, resetting preferences, clearing local tasks and memory, or deleting the account; export local data before destructive actions. ![Uninstalling and resetting Foxora in Foxora v5.0.3](/docs/images/v5.0.3/settings/data.webp) ## Where this lives Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. Foxora v5.0.3 has two top-level workspaces: **foxora** for tasks, projects, workflows, and schedules; and **studio** for bots, agents, crews, skills, tools, memory, and plugins. > **Safety:** Download installers only from Foxora or the official release repository. Attach only folders needed for the task. ## Steps 1. Start in the Foxora sign-in screen, **New Task**, or **Settings → About**. 2. Choose between removing the application, resetting preferences, clearing local tasks and memory, or deleting the account. 3. Export local data before destructive actions. 4. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 5. **Confirm the result.** Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. ## Choose the right path - Use Desktop for local projects and visual work. - Use the web app when you do not need local files. - Use the CLI for terminal-first or scripted work. ## Confirm it worked - Create a task from **New Task**, send a small request, and confirm the result in the transcript, Stage, project, or external destination. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/overview title: Overview --- # Overview Learn the workspace regions and the rule that the active task controls the conversation, project, crew, autonomy mode, and Stage context. ![Foxora Workspace in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Learn the workspace regions and the rule that the active task controls the conversation, project, crew, autonomy mode, and Stage context. 3. Choose the active task and project first, then use the visible header actions for Runs, transcript view, and the right panel. 4. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/application-layout-and-navigation title: Application layout and navigation --- # Application layout and navigation Move between the task rail, conversation, composer, Studio, Settings, and Stage without losing the active task. ![Application layout and navigation in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Move between the task rail, conversation, composer, Studio, Settings, and Stage without losing the active task. 3. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/command-center title: Command Center --- # Command Center Open the Command Center with Ctrl+K or Command+K, search tasks and actions, use keyboard navigation, and execute the highlighted result. ![Command Center in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/command-center.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Open the Command Center with Ctrl+K or Command+K. 3. Search tasks and actions. 4. Use keyboard navigation, and execute the highlighted result. 5. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/panels-and-saved-view-state title: Panels and saved view state --- # Panels and saved view state Resize, collapse, and reopen workspace panels; confirm Foxora restores useful layout state without changing task data. ![Panels and saved view state in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Resize, collapse, and reopen workspace panels. 3. Confirm Foxora restores useful layout state without changing task data. 4. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/keyboard-shortcuts title: Keyboard shortcuts --- # Keyboard shortcuts Open the shortcut reference, learn navigation and composer shortcuts, and resolve conflicts with operating-system or editor bindings. ![Keyboard shortcuts in Foxora v5.0.3](/docs/images/v5.0.3/settings/keyboard.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Open the shortcut reference, learn navigation and composer shortcuts. 3. Resolve conflicts with operating-system or editor bindings. 4. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workspace/appearance-and-themes title: Appearance and themes --- # Appearance and themes Choose the color scheme, accent, density, and motion preferences; verify readability in conversation, editor, terminal, and Stage surfaces. ![Appearance and themes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/settings-appearance.webp) ## Where this lives Open the **foxora** workspace and keep the left sidebar visible. The left sidebar contains **New Task**, **Projects**, **Workflows**, and **Schedules**, followed by project and task lists. The center holds the active task and composer; the right side holds Stage or Runs. > **Safety:** Panel layout is presentation state. Closing a panel should not be treated as stopping an active agent or terminal process. ## Steps 1. Open the **foxora** workspace and keep the left sidebar visible. 2. Choose the color scheme, accent, density, and motion preferences. 3. Verify readability in conversation, editor, terminal, and Stage surfaces. 4. **Confirm the result.** Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. ## Choose the right path - Collapse the Stage for conversation-focused work. - Open the editor or terminal for builder workflows. - Use the Command Center for fast navigation and actions. ## Confirm it worked - Switch between Tasks, Projects, Workflows, and Schedules and confirm that each destination opens without losing the active composer or task context. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/overview title: Overview --- # Overview Understand what belongs to a task and how it differs from a project, agent, crew, workflow, schedule, and run. ![Tasks in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/tasks-overview.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Understand what belongs to a task and how it differs from a project, agent, crew, workflow, schedule. 3. Run. 4. Create one task for one outcome, attach only the required project, select the crew and autonomy mode, and keep unrelated work in separate tasks. 5. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Local runtime boundary | Surface | Session and Memory Den behavior | | --- | --- | | Foxora Desktop | Reads and writes the device’s local Foxora runtime. | | Foxora CLI | Connects to the same local runtime as Desktop on that device. | | Foxora web app | Does not automatically receive the device’s local sessions or memories. | | Another device | Has its own local runtime unless data is deliberately exported and moved. | ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/starting-a-session title: Starting a task --- # Starting a task Choose New Task, select a project when file work is required, set the crew and autonomy mode, and send an outcome-focused first request. ![Starting a task in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Choose New Task. 3. Select a project when file work is required. 4. Set the crew and autonomy mode. 5. Send an outcome-focused first request. 6. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/managing-sessions title: Managing tasks --- # Managing tasks Use the Tasks screen, search, filters, grid or list view, status badges, and Open action to find and resume work. ![Managing tasks in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/tasks-overview.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Use the Tasks screen. 3. Search, filters, grid or list view, status badges. 4. Open action to find and resume work. 5. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/renaming-pinning-and-archiving-sessions title: Renaming and archiving tasks --- # Renaming and archiving tasks Use the task action menu to give work an outcome-based name, archive completed items, and restore them when the work becomes active again. ![Renaming and archiving tasks in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/tasks-overview.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Use the task action menu to give work an outcome-based name, archive completed items, and restore them when the work becomes active again. 3. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/forking-sessions title: Branching from an existing task --- # Branching from an existing task Create a separate task before trying a substantially different approach so the original transcript and result remain reviewable. ![Branching from an existing task in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Create a separate task before trying a substantially different approach so the original transcript and result remain reviewable. 3. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/checkpoints-and-session-history title: Checkpoints and task history --- # Checkpoints and task history Use the transcript and task state to locate important transitions, preserve evidence, and recover only after reviewing what will change. ![Checkpoints and task history in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Use the transcript and task state to locate important transitions. 3. Preserve evidence, and recover only after reviewing what will change. 4. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/sessions/multi-agent-sessions title: Multi-agent tasks --- # Multi-agent tasks Select Vixen, Zenko, or Kyubi Crew, follow delegated agent work, and verify the integrated result delivered by the lead. ![Multi-agent tasks in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. A task is the user-facing unit of work in v5.0.3. It has a transcript, project context, selected crew, autonomy mode, run state, and Stage surfaces. Internal APIs may still use a task identifier. > **Safety:** Task transcripts may include local files, tool results, and connected-app data. Review before sharing, exporting, or attaching them to support requests. ## Steps 1. Open **foxora → Tasks** for the collection view, or choose a task in the sidebar. 2. Select Vixen, Zenko, or Kyubi Crew. 3. Follow delegated agent work. 4. Verify the integrated result delivered by the lead. 5. **Confirm the result.** Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. ## Choose the right path - Start a clean task for a new objective. - Branch from an existing task when you need its context but want a different approach. - Archive completed work without deleting its underlying project. ## Confirm it worked - Reopen the task from the Tasks list and confirm its transcript, project, crew, mode, and latest result are intact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/overview title: Overview --- # Overview Understand how projects provide file boundaries and working directories while tasks provide conversational history and task state. ![Projects in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Understand how projects provide file boundaries and working directories while tasks provide conversational history and task state. 3. Open the narrowest appropriate folder, confirm the active working directory, review repository state, and add project instructions for durable conventions. 4. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/creating-and-opening-projects title: Creating and opening projects --- # Creating and opening projects Create a new folder or select an existing one, review the resolved path, attach it to the task, and confirm it becomes active. ![Creating and opening projects in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Create a new folder or select an existing one. 3. Review the resolved path. 4. Attach it to the task. 5. Confirm it becomes active. 6. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/working-with-local-folders title: Working with local folders --- # Working with local folders Open a local folder, inspect its top-level files, exclude unrelated or sensitive locations, and run a read-only orientation task first. ![Working with local folders in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Open a local folder. 3. Inspect its top-level files. 4. Exclude unrelated or sensitive locations. 5. Run a read-only orientation task first. 6. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/working-with-git-repositories title: Working with Git repositories --- # Working with Git repositories Open or clone a repository, check branch and working-tree status, read repository instructions, and avoid mixing unrelated changes. ![Working with Git repositories in Foxora v5.0.3](/docs/images/v5.0.3/workspace/new-project.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Open or clone a repository. 3. Check branch and working-tree status. 4. Read repository instructions, and avoid mixing unrelated changes. 5. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/project-context-and-instructions title: Project context and instructions --- # Project context and instructions Add durable project rules such as commands, formatting, architecture, and verification expectations; test that a new task follows them. ![Project context and instructions in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Add durable project rules such as commands, formatting, architecture, and verification expectations. 3. Test that a new task follows them. 4. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/scratchpads-and-temporary-workspaces title: Scratchpads and temporary workspaces --- # Scratchpads and temporary workspaces Start without a folder for questions or planning, then attach or create a project only when the task needs persistent files. ![Scratchpads and temporary workspaces in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Start without a folder for questions or planning. 3. Attach or create a project only when the task needs persistent files. 4. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/projects/managing-and-removing-projects title: Managing and removing projects --- # Managing and removing projects Change the active project, detach a folder from a task, remove it from recent projects, and distinguish that from deleting files on disk. ![Managing and removing projects in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/projects-overview.webp) ## Where this lives Open **foxora → Projects** and select the project you intend to use. A project grants a task access to a specific local folder or Git repository and supplies the working directory used by files, shell commands, and project instructions. > **Safety:** Foxora should not receive a parent folder merely for convenience. A project can contain secrets even when they are not mentioned in the prompt. ## Steps 1. Open **foxora → Projects** and select the project you intend to use. 2. Change the active project, detach a folder from a task. 3. Remove it from recent projects. 4. Distinguish that from deleting files on disk. 5. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 6. **Confirm the result.** Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. ## Choose the right path - Use a local folder for existing work. - Clone or open a Git repository when history matters. - Use a scratchpad when no file access is required. ## Confirm it worked - Ask Foxora to identify the active project, read a harmless file, and show the exact path without changing anything. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/web-cloud/overview title: Overview --- # Web App & Cloud overview The Foxora web app brings tasks, projects, crews, runs, and generated artifacts into a browser. When a web task needs files or commands, Foxora gives the signed-in user an isolated Foxora Cloud workspace instead of reaching into the computer that opened the page. ![Web App & Cloud in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/new-task.webp) ## What the web app includes | Area | What happens in the web app | | --- | --- | | Tasks | Create and resume browser tasks, choose a crew and autonomy mode, and review the transcript. | | Projects | Work in cloud project folders or the task’s generated scratch project. | | Files and editor | Browse, create, edit, rename, and download files stored in Foxora Cloud. | | Terminal | Run a real cloud shell for the active project. Terminal tabs can reconnect with bounded scrollback while the browser task remains open. | | Artifacts | Review generated Markdown, code, diagrams, tables, and other supported output in Preview or Code. | | Runs | Follow background work without assuming that closing the Runs panel cancels it. | Foxora Cloud is implemented with an isolated Daytona-backed workspace. Foxora owns the workspace identity, path mapping, resume behavior, and backup lifecycle; the product UI refers to the result as a Foxora Cloud workspace. ## What does not cross over automatically - A folder opened in Desktop is not silently uploaded to the web app. - Desktop and CLI task history stored only on a device does not automatically become web task history. - A cloud project does not appear as a local folder until you download it or deliberately copy it into a local project. - Local browser profiles, signed-in desktop applications, and device permissions remain on the device. Use Desktop when the outcome depends on the current computer. Use the web app when the work should run in an isolated browser-accessible workspace. ## Complete a first web check 1. Open [app.foxora.ai](https://app.foxora.ai/) and sign in. 2. Choose **New Task**. 3. Read the selected autonomy mode and crew before sending anything. 4. Ask for one small output, such as a short Markdown comparison or a tiny project with a README. 5. Expand the tool activity if the task uses the cloud workspace. 6. Inspect the final artifact in both **Preview** and **Code** when both views are available. 7. Download or copy the output if it must exist outside Foxora Cloud. The web runtime is ready when the task completes, the transcript shows the expected tool activity, and the final result opens or downloads successfully. --- source: https://docs.foxora.ai/web-cloud/starting-a-web-task title: Starting a web task --- # Starting a web task A web task starts from the same outcome-first composer as Desktop, but any filesystem or command work runs inside Foxora Cloud. ![Starting a web task in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/new-task.webp) ## Start with the controls that affect the run 1. Open [app.foxora.ai](https://app.foxora.ai/) and choose **New Task**. 2. Confirm the account and credit balance shown in the workspace. 3. Open the autonomy pill below the composer: - Use **Ask Permissions** when you want each mutating step reviewed. - Use **Accept Edits** when routine file edits may proceed but riskier commands should remain gated. - Use **Auto mode** for a bounded, recoverable cloud task. - Reserve **Bypass Permissions** for an isolated project whose contents and external access are understood. 4. Open the crew picker and choose Vixen, Zenko, or Kyubi Crew according to the available plan and required depth. 5. Describe the outcome, constraints, and evidence required to call it complete. 6. Attach only the files needed for this task, then send. ## Write a request the workspace can verify Prefer a request such as: > Build a small responsive product-status page in this cloud project. Keep dependencies minimal, run the build, and show me the final page plus the exact verification result. This gives Foxora an output, a scope, and a test. Avoid requests such as “make something nice,” which provide no stable completion signal. ## Understand the generated project If no project is attached, Foxora gives the task its own scratch project folder. File tools, shell commands, the Files tree, and project download resolve to that same folder so generated work does not drift into a shared workspace root. The project appears in **foxora → Projects** after the task has a cloud project to show. Give it a useful name once the experiment becomes durable work. ## Follow the task - Use **Focused** transcript view for the main decisions and results. - Use **Verbose** when diagnosing tool arguments, command output, or a failed step. - Open **Runs** to see background work; closing the panel only closes the view. - Interrupt the task if it targets the wrong project, account, or external destination. The task is complete only after the real artifact, file, command result, preview, or external destination matches the acceptance criteria. --- source: https://docs.foxora.ai/web-cloud/cloud-projects-and-workspaces title: Cloud projects and workspaces --- # Cloud projects and workspaces A Foxora Cloud workspace belongs to the signed-in user. Projects are folders inside that isolated workspace, while tasks hold the conversation, crew, autonomy mode, run state, and artifact history. ![Cloud projects and workspaces in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/projects.webp) ## Project and task are different | Object | Contains | What happens when you switch it | | --- | --- | --- | | Task | Transcript, selected crew, autonomy mode, runs, and artifacts | Reopens the recorded work and its current context. | | Project | Source files, folders, Git history, and generated project files | Changes the filesystem and command working directory. | | Cloud workspace | All cloud project folders for the signed-in user | Is created or resumed by Foxora when cloud tools need it. | A task without a user-selected project receives a generated per-task scratch project. That keeps its shell commands, file writes, Files tree, and downloads anchored to one location. ## Work with the project collection 1. Open **foxora → Projects**. 2. Search by the project name or generated folder identifier. 3. Use grid or list view to inspect the collection. 4. Select **Open** on the intended project. 5. Confirm the project name in the task header before asking Foxora to edit or run anything. The collection shows cloud projects, not folders from the computer running the browser. ## Create, import, and export - **New project** creates a folder in Foxora Cloud. - A pasted public Git URL can be cloned into the cloud workspace. - A repository selected through a connected GitHub account uses the authorized import path; private credentials are not copied into the sandbox. - A single file can be downloaded directly from the Files view. - A folder or complete project can be packaged server-side and downloaded as an archive. Downloading is the boundary that moves project bytes from Foxora Cloud to the current computer. Confirm the browser completed the download before removing or replacing a cloud project. ## Keep projects reviewable - Use one project for one codebase or durable body of work. - Rename generated scratch projects that become long-lived. - Keep secrets out of project files where possible; files in the project may be included in workspace backups. - Preserve Git history and run the project’s own tests before exporting or publishing it. - Archive a finished task without assuming that the underlying project folder has been deleted. --- source: https://docs.foxora.ai/web-cloud/cloud-execution-and-artifacts title: Cloud execution and artifacts --- # Cloud execution and artifacts Cloud execution is visible in the task transcript. Foxora shows reasoning summaries, tool activity, commands, and generated artifacts in the same conversation so the result can be reviewed before it leaves the cloud project. ![Cloud execution and artifacts in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/cloud-artifact.webp) ## Read the execution record 1. Open the task from the sidebar or **foxora → Tasks**. 2. Use **Focused** transcript view for the main path. 3. Switch to **Verbose** when you need exact tool and command evidence. 4. Expand a **Ran**, **Read**, **Wrote**, or other tool row to inspect its target and result. 5. Check that commands ran in the intended cloud project rather than an unrelated folder. 6. Treat a green or completed tool row as evidence for that action, not proof that the whole outcome is correct. The cloud shell starts in the active project. Each tool command may run in a fresh shell, so a standalone `cd` should not be treated as durable state; Foxora anchors later commands to the project again. ## Review an artifact Supported artifacts appear inline and can also open on the Stage: - **Preview** renders the artifact when Foxora has a renderer for its type. - **Code** shows the underlying source. - **Copy** uses the artifact source or relevant URL. - **Download** saves a supported artifact or file to the browser’s download location. For a report, verify the facts and links. For code, run the build or tests. For a visual, inspect the rendered result at the required viewport. For a file, download it and open the downloaded copy. ## Recognize an incomplete result Do not accept the task as complete when: - the final reply only describes work that the transcript never performed; - the artifact renders but its source is truncated or invalid; - a command succeeded in the wrong project; - a development server started but the page was never opened or checked; - an external action is claimed without a destination URL, identifier, or visible final state. Ask Foxora to perform the missing verification in the same task while the project context is still active. --- source: https://docs.foxora.ai/web-cloud/desktop-versus-web title: Desktop versus web --- # Desktop versus web Desktop and web share Foxora’s task-oriented interface, but they execute against different machines. Choose the surface from where the required files, applications, and account state actually live. ![Desktop versus web in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/new-task.webp) | Capability | Desktop | Web app | | --- | --- | --- | | Working machine | This computer | Isolated Foxora Cloud workspace | | Project files | Folders you explicitly open on the device | Cloud project folders | | Editor and terminal | Local project and local shell | Cloud project and cloud shell | | Desktop applications | Available through approved Computer Use permissions | Not available | | Existing browser profile | Available when explicitly authorized | Uses browser-accessible or managed cloud behavior; it does not inherit Desktop’s local profile | | Long-running workspace | Depends on the local app and runtime | Cloud machine can suspend and resume independently of the laptop | | Moving files to the other surface | Open or copy them locally | Download files or a project archive | ## Choose Desktop when - the task must edit a folder already on this computer; - Foxora needs a local Git worktree, local development service, or installed application; - the task needs an authenticated local browser profile; - you need native file pickers, desktop notifications, microphone, Accessibility, or Screen Recording. ## Choose web when - you want to begin from any supported browser without installing the app; - the work should stay in an isolated cloud project; - a cloud shell and web editor are sufficient; - the task should remain available after the current browser or laptop is closed. ## Avoid false continuity Using the same account does not mean every local object is synchronized: - Desktop and CLI on the same device can use the same local runtime and data home. - The web app uses cloud task and workspace state. - Local folders remain local until you intentionally upload, clone, or copy their contents. - Cloud output remains cloud-side until you download it or push it to an external repository. Before continuing a task on another surface, confirm the project files, branch, task transcript, and latest artifact are actually present there. --- source: https://docs.foxora.ai/web-cloud/cloud-workspace-lifecycle title: Cloud workspace lifecycle --- # Cloud workspace lifecycle Foxora creates one cloud machine for the signed-in user and stores projects as folders inside it. The machine starts or resumes only when the web task needs cloud execution, then uses suspension and backup policies to control cost without treating the machine as the only durable copy. ![Cloud workspace lifecycle in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/projects.webp) ## Current v5.0.3 lifecycle 1. **Warm or provision.** Foxora begins warming the user’s cloud machine after the web task connects. First use creates it when no machine exists. 2. **Run.** File tools, the editor, terminal, and task commands resolve under the Foxora Cloud workspace root. 3. **Back up changes.** File operations, completed turns, clones, and terminal activity schedule a coalesced project snapshot instead of uploading after every keystroke. 4. **Suspend when idle.** The cloud machine can auto-stop after approximately 15 minutes of inactivity. Suspension preserves the machine and is expected behavior. 5. **Resume on demand.** A later operation refreshes stale provider state and starts a stopped machine before accessing files or the terminal. 6. **Recreate when required.** When durable backup is enabled, a machine may be removed after a longer idle period and recreated later. Foxora restores the project snapshot into the replacement. Task and memory records for the web surface live in managed storage. Project files and Git history are backed up separately from the cloud machine’s local disk. ## What a project backup contains The project snapshot includes source files and Git metadata. Large or derived directories such as `node_modules`, build output, caches, Rust `target`, and Python bytecode are excluded because they can be rebuilt from source and manifests. After a restore, reinstall dependencies and rerun the project’s build or tests before relying on previously generated output. ## What users should do - Keep source and dependency manifests inside the project folder. - Commit important milestones to Git. - Download irreplaceable deliverables before deleting a task or project. - Do not store secrets in ordinary project files unless the task truly requires them. - Expect a cold resume to take longer than a warm command; do not create duplicate tasks merely because the first cloud operation is starting. The correct continuity test is not whether the machine kept the same internal identifier. Reopen the task, open the project, read a known source file, and run the smallest project-specific verification. --- source: https://docs.foxora.ai/web-cloud/troubleshooting-cloud-workspaces title: Troubleshooting cloud workspaces --- # Troubleshooting cloud workspaces Most web workspace failures fall into one of four layers: browser task, task/project binding, cloud machine readiness, or the command itself. Diagnose them in that order and preserve the existing project before creating replacements. ![Troubleshooting cloud workspaces in Foxora v5.0.3](/docs/images/v5.0.3/web-cloud/cloud-execution.webp) ## First checks 1. Confirm [app.foxora.ai](https://app.foxora.ai/) still shows the expected account. 2. Reopen the task from **foxora → Tasks**. 3. Confirm the intended project name in the header and **foxora → Projects**. 4. Switch the transcript to **Verbose** and capture the exact failed tool row. 5. Retry one read-only action, such as listing the project root or reading a known file. Do not delete the project, clear browser data, or start multiple replacement tasks before this check. ## Match the symptom | Symptom | Likely layer | Safe next action | | --- | --- | --- | | First file or terminal action is slow | Cold cloud-machine start | Wait for the current action, then retry one read-only operation once. | | Workspace or terminal reports unavailable | Cloud readiness or expired browser task | Refresh after preserving unsent text, sign in again if prompted, and reopen the same task. | | Files appear in the transcript but not the Files tree | Wrong active project or stale view | Check the project header, reopen the project, then refresh the Files view. | | Export says there is nothing to download | Task has no output in the selected project | Verify the task’s generated scratch project and ask Foxora to name the exact file path. | | Command remains running | Interactive prompt or long process | Inspect the command, interrupt it if safe, then rerun with non-interactive flags or a bounded timeout. | | A recreated project lacks dependencies | Derived folders were intentionally excluded from backup | Reinstall from the lockfile and rerun the build. | | Artifact preview fails | Renderer or invalid artifact source | Switch to **Code**, download the source, and inspect the error before regenerating. | ## Recover without losing evidence 1. Copy the exact error, task name, project name, and approximate time. 2. Download any still-accessible artifact or project archive. 3. Ask Foxora to summarize completed steps and unresolved work in the current task. 4. Retry the smallest failing operation. 5. Create a new task only if the existing task cannot reconnect, and attach the existing project rather than generating another empty one. If the same read-only operation fails after a refresh and reauthentication, report the browser, Foxora web version, task identifier, project name, failed tool, and timestamp. Do not include access tokens, provider identifiers, private environment variables, or unrelated project files. --- source: https://docs.foxora.ai/composer/overview title: Overview --- # Overview Learn how requests, attachments, slash commands, crew selection, autonomy, and follow-up messages combine within one task. ![Task Composer in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Learn how requests, attachments, slash commands, crew selection, autonomy. 3. Follow-up messages combine within one task. 4. Describe the outcome and constraints, attach only relevant material, choose Vixen, Zenko, or Kyubi Crew, and select the least permissive autonomy mode that will complete the work. 5. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/sending-your-first-request title: Sending your first request --- # Sending your first request Write one outcome-focused request, include constraints and verification criteria, send it, and review both the answer and performed actions. ![Sending your first request in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Write one outcome-focused request, include constraints and verification criteria. 3. Send it. 4. Review both the answer and performed actions. 5. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/slash-commands title: Slash commands --- # Slash commands Type `/` to browse the built-in commands, choose the workflow that matches the task, add arguments, and review the expanded work in the transcript. ![Slash commands in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/command-center.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Type `/` to browse the built-in commands. 3. Choose the workflow that matches the task. 4. Add arguments. 5. Review the expanded work in the transcript. 6. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/adding-files-images-and-attachments title: Adding files, images, and attachments --- # Adding files, images, and attachments Attach a file or image from the composer, confirm the correct item and size, explain what to inspect, and remove accidental attachments before sending. ![Adding files, images, and attachments in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/new-task.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Attach a file or image from the composer. 3. Confirm the correct item and size, explain what to inspect. 4. Remove accidental attachments before sending. 5. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/selecting-a-crew-or-processing-tier title: Selecting a crew or processing tier --- # Selecting a crew or processing tier Open the crew picker, choose Fast, Pro, or Max based on complexity and plan access, then verify the selected crew before sending. The Pro crew uses Foxora’s premium processing pool. ![Selecting a crew or processing tier in Foxora v5.0.3](/docs/images/v5.0.3/workspace/crew-picker.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Open the crew picker. 3. Choose Fast, Pro, or Max based on complexity and plan access. 4. Verify the selected crew before sending. The Pro crew uses Foxora’s premium processing pool. 5. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/voice-and-hands-free-conversations title: Voice and hands-free conversations --- # Voice and hands-free conversations Enable microphone access, select input and output devices, dictate a concise request, review the transcription, and stop listening when finished. ![Voice and hands-free conversations in Foxora v5.0.3](/docs/images/v5.0.3/settings/voice.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Enable microphone access. 3. Select input and output devices, dictate a concise request. 4. Review the transcription. 5. Stop listening when finished. 6. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/context-and-compaction title: Context and compaction --- # Context and compaction Keep the task focused, ask for a summary before long transitions, inspect compaction notices, and start or fork a task when the objective changes. ![Context and compaction in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open **New Task** or an existing task and focus the composer. The v5.0.3 composer combines the prompt, Enhance, attachments, voice, autonomy, crew selection, credit status, starter actions, and Send. > **Safety:** Do not paste passwords, API keys, or unrelated sensitive content into a message. Use dedicated connection or secret controls. ## Steps 1. Open **New Task** or an existing task and focus the composer. 2. Keep the task focused, ask for a summary before long transitions. 3. Inspect compaction notices. 4. Start or fork a task when the objective changes. 5. **Confirm the result.** Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. ## Choose the right path - Send a normal request for flexible work. - Use a slash command for a known workflow. - Attach files or images when the agent must inspect specific source material. ## Confirm it worked - Read the transcript and tool activity, then inspect the actual file, Stage surface, workflow, schedule, or external destination produced by the task. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/composer/transcripts-and-conversation-history title: Transcripts and task history --- # Transcripts and conversation history Review the local record of a task, choose how much activity to display, and preserve important evidence before removing local data. ## Before you start - A transcript belongs to its task and is stored locally on the device running the Foxora runtime. - Desktop and CLI can access the same local tasks when they connect to that runtime. The web app does not automatically synchronize local transcripts to another device. - Do not copy or export secrets, private prompts, or sensitive tool output unless the destination is approved. ## Open the transcript controls 1. Select the task you want to inspect from the left sidebar. 2. Select the transcript menu in the upper-right corner of the conversation. 3. Choose the presentation that fits the review: - **Focused** keeps the conversation easier to scan. - **Verbose** exposes more operational detail when you need to audit agent and tool activity. ![Transcripts and task history in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) Changing this setting changes how the transcript is presented; it does not create a new task or alter the underlying project files. ## Review a task step by step 1. Start at the request that produced the result you are checking. 2. Read the response and expand any agent, tool, command, or approval activity relevant to the result. 3. Match referenced files, commands, browser actions, or artifacts with their real destination. 4. Copy only the portion needed for review or support. 5. Remove secrets and personal or proprietary information before sharing copied content. 6. Verify the final state in the project, terminal, browser, Stage, or connected service instead of relying only on the conversation summary. ## If you want to… - Scan normal conversation quickly, use **Focused**. - Investigate how a result was produced, switch to **Verbose**. - Compare two approaches without losing the original history, fork the task before continuing. - Preserve important history before a reset, export or copy the required evidence first. ## What success looks like - You are reviewing the intended task and can identify the request that produced the result. - The transcript provides enough detail to connect the response with the actions Foxora performed. - The actual file, process, browser page, artifact, or external destination confirms the reported outcome. --- source: https://docs.foxora.ai/autonomy-approvals/overview title: Overview --- # Overview Understand the four task autonomy modes, their approval boundaries, and the separate `/plan` command. ![Autonomy & Approvals in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Understand the four task autonomy modes, their approval boundaries, and the separate `/plan` command. 3. Start with the lowest mode that permits the workflow, read each approval request, and raise autonomy only after the task and environment are understood. 4. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Mode comparison | Mode | Best for | Expected behavior | | --- | --- | --- | | Ask Permissions | Unfamiliar, sensitive, or high-impact work | Pauses for approval before actions that need permission. | | Accept Edits | Supervised building and coding | Allows routine edits while retaining gates for commands and riskier actions. | | Auto | Trusted, bounded execution | Proceeds independently within the available tools and permissions. | | Bypass Permissions | Disposable or tightly controlled environments | Removes normal approval pauses; external effects remain real and may be irreversible. | > **Plan is not an autonomy mode.** Use `/plan` when you want Foxora to outline and track the work before acting. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/ask-permissions-mode title: Ask Permissions mode --- # Ask Permissions mode Select Ask Permissions, send a task that needs a file edit or command, inspect the approval details, and approve or decline each action. ![Ask Permissions mode in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Select Ask Permissions. 3. Send a task that needs a file edit or command. 4. Inspect the approval details. 5. Approve or decline each action. 6. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/accept-edits-mode title: Accept Edits mode --- # Accept Edits mode Allow routine file edits while keeping command and higher-risk actions gated; review diffs frequently during the task. ![Accept Edits mode in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Allow routine file edits while keeping command and higher-risk actions gated. 3. Review diffs frequently during the task. 4. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/auto-mode title: Auto mode --- # Auto mode Use Auto for a bounded task with clear constraints, monitor the transcript, and interrupt when Foxora reaches an unexpected decision or external side effect. ![Auto mode in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Use Auto for a bounded task with clear constraints. 3. Monitor the transcript. 4. Interrupt when Foxora reaches an unexpected decision or external side effect. 5. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/bypass-permissions-mode title: Bypass Permissions mode --- # Bypass Permissions mode Confirm the project is trusted and recoverable, select Bypass Permissions, acknowledge the warning, and return to a safer mode immediately afterward. ![Bypass Permissions mode in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Confirm the project is trusted and recoverable. 3. Select Bypass Permissions. 4. Acknowledge the warning, and return to a safer mode immediately afterward. 5. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/reviewing-approval-requests title: Reviewing approval requests --- # Reviewing approval requests Read the action, target, command, working directory, and risk; approve once, decline, or revise the request instead of approving unclear work. ![Reviewing approval requests in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Read the action, target, command, working directory, and risk. 3. Approve once, decline, or revise the request instead of approving unclear work. 4. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/changing-autonomy-during-a-task title: Changing autonomy during a task --- # Changing autonomy during a task Pause or wait for a safe boundary, change the task pill, confirm the new mode in the composer, and continue with an explicit instruction. ![Changing autonomy during a task in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Pause or wait for a safe boundary. 3. Change the task pill. 4. Confirm the new mode in the composer, and continue with an explicit instruction. 5. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/autonomy-approvals/choosing-the-appropriate-autonomy-mode title: Choosing the appropriate autonomy mode --- # Choosing the appropriate autonomy mode Classify the task by reversibility, data sensitivity, external impact, and environment; then choose the lowest mode that avoids unnecessary interruptions. ![Choosing the appropriate autonomy mode in Foxora v5.0.3](/docs/images/v5.0.3/workspace/autonomy-picker.webp) ## Where this lives Open the autonomy control beside the composer. Autonomy is set per task: Ask Permissions, Accept Edits, Auto, or Bypass Permissions. It controls when Foxora pauses before file, command, and consequential actions. > **Safety:** Autonomy does not make external actions reversible. Publishing, sending, purchasing, deleting, and production changes require deliberate review. ## Steps 1. Open the autonomy control beside the composer. 2. Classify the task by reversibility, data sensitivity, external impact, and environment. 3. Choose the lowest mode that avoids unnecessary interruptions. 4. **Confirm the result.** Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. ## Choose the right path - Ask Permissions for unfamiliar or sensitive work. - Accept Edits for supervised coding. - Auto for trusted, bounded tasks. - Bypass Permissions only in disposable or tightly controlled environments. ## Confirm it worked - Confirm that safe actions proceed at the intended level and that gated actions stop with a clear approval card. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/overview title: Overview --- # Overview Understand how the Stage follows the active task and switches between plans, artifacts, files, terminal, browser, media, and supporting views. ![The Stage in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-switcher.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Understand how the Stage follows the active task and switches between plans, artifacts, files, terminal, browser, media, and supporting views. 3. Keep the active task selected, focus the relevant Stage surface, and use its controls to inspect rather than infer progress. 4. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/understanding-stage-panels title: Understanding Stage panels --- # Understanding Stage panels Open a Stage item from the transcript, switch surfaces from the Stage menu, close finished surfaces, resize the panel, and return to the conversation without losing the active surface. ![Understanding Stage panels in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-switcher.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Open a Stage item from the transcript. 3. Switch surfaces from the Stage menu. 4. Close finished surfaces, resize the panel, and return to the conversation without losing the active surface. 5. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/plans-and-live-progress title: Plans and live progress --- # Plans and live progress Open the plan checklist, follow active and completed steps, inspect blocked items, and verify the final step against real output. ![Plans and live progress in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Open the plan checklist. 3. Follow active and completed steps. 4. Inspect blocked items. 5. Verify the final step against real output. 6. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/artifacts-and-generated-files title: Artifacts and generated files --- # Artifacts and generated files Open an artifact from the transcript, inspect its rendered and source views, copy or save it, and confirm whether it is also written to the project. ![Artifacts and generated files in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Open an artifact from the transcript. 3. Inspect its rendered and source views. 4. Copy or save it. 5. Confirm whether it is also written to the project. 6. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/images-diagrams-tables-and-media title: Images, diagrams, tables, and media --- # Images, diagrams, tables, and media Open the generated media surface, zoom or switch representation, check labels and underlying data, and export in the required format. ![Images, diagrams, tables, and media in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Open the generated media surface, zoom or switch representation. 3. Check labels and underlying data. 4. Export in the required format. 5. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/stage/supporting-surfaces title: Supporting surfaces --- # Supporting surfaces Use process, memory, source-control, terminal, or browser panels to inspect the evidence behind an agent response and take over when needed. ![Supporting surfaces in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-switcher.webp) ## Where this lives Open a task, reveal the right panel, and use the Stage title to open its surface switcher. The Stage presents the live surface relevant to the current work: plan, artifact, file, terminal, browser, source-control view, media, process, or recalled memory. > **Safety:** A rendered preview is evidence, not always the source of truth. Verify persistent output in its actual destination. ## Steps 1. Open a task, reveal the right panel, and use the Stage title to open its surface switcher. 2. Use process, memory, source-control, terminal, or browser panels to inspect the evidence behind an agent response and take over when needed. 3. **Confirm the result.** Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. ## Choose the right path - Follow a plan for multi-step work. - Open an artifact for a polished output. - Use supporting surfaces to inspect terminals, files, browser state, or memory. ## Confirm it worked - Confirm the Stage result agrees with the transcript and the underlying file, process, browser page, or artifact. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/overview title: Overview --- # Overview Learn how project-scoped terminals, streamed output, approvals, background processes, and exit codes appear in Foxora. ![Terminal in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Learn how project-scoped terminals, streamed output, approvals, background processes, and exit codes appear in Foxora. 3. Confirm the working directory and command, choose an appropriate autonomy mode, then run the smallest command that proves the next step. 4. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/running-commands title: Running commands --- # Running commands Open a terminal in the active project, inspect the command and working directory, approve when required, and read stdout, stderr, and exit code. ![Running commands in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Open a terminal in the active project. 3. Inspect the command and working directory. 4. Approve when required. 5. Read stdout, stderr, and exit code. 6. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/interactive-processes title: Interactive processes --- # Interactive processes Start an interactive command, keep its terminal focused for input, respond to prompts deliberately, and stop it cleanly when finished. ![Interactive processes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Start an interactive command. 3. Keep its terminal focused for input, respond to prompts deliberately. 4. Stop it cleanly when finished. 5. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/managing-multiple-terminals title: Managing multiple terminals --- # Managing multiple terminals Create separate terminals for independent processes, name or identify each by command, switch without mixing input, and close only completed processes. ![Managing multiple terminals in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Create separate terminals for independent processes. 3. Name or identify each by command. 4. Switch without mixing input. 5. Close only completed processes. 6. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/stopping-and-restarting-processes title: Stopping and restarting processes --- # Stopping and restarting processes Send the normal stop signal first, confirm the process exited and released its port, then restart with the same verified configuration. ![Stopping and restarting processes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Send the normal stop signal first. 3. Confirm the process exited and released its port. 4. Restart with the same verified configuration. 5. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 6. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/terminal/terminal-permissions-and-safety title: Terminal permissions and safety --- # Terminal permissions and safety Inspect commands before approval, check path and environment assumptions, avoid untrusted shell interpolation, and use least privilege. ![Terminal permissions and safety in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-terminal.webp) ## Where this lives Open a project-backed task, then choose **Stage → Terminal**. Terminals run commands in the active project working directory and stream stdout, stderr, process state, and exit status into an inspectable surface. > **Safety:** Review commands that install software, change Git state, access secrets, use elevated privileges, or affect files outside the project. ## Steps 1. Open a project-backed task, then choose **Stage → Terminal**. 2. Inspect commands before approval. 3. Check path and environment assumptions, avoid untrusted shell interpolation. 4. Use least privilege. 5. **Confirm the result.** Check the exit code and resulting system or file state; successful-looking output alone is not enough. ## Choose the right path - Run a foreground command for short tasks. - Use a background process for servers and watchers. - Open separate terminals for independent long-running processes. ## Confirm it worked - Check the exit code and resulting system or file state; successful-looking output alone is not enough. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/overview title: Overview --- # Overview Understand the project tree, editor tabs, dirty state, agent changes, generated files, and the relationship to source control. ![Files & Editor in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-files.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Understand the project tree, editor tabs, dirty state, agent changes, generated files, and the relationship to source control. 3. Confirm the active project and file path, open the relevant file, make or review a focused change, then save and validate it with the project’s normal checks. 4. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/browsing-project-files title: Browsing project files --- # Browsing project files Expand folders, search for a file, open it in a tab, confirm its path and language, and use symbols or content search to navigate. ![Browsing project files in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-files.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Expand folders. 3. Search for a file. 4. Open it in a tab. 5. Confirm its path and language. 6. Use symbols or content search to navigate. 7. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/creating-and-editing-files title: Creating and editing files --- # Creating and editing files Create a file in the intended folder, enter or request the content, save it, and run the project-specific verification command. ![Creating and editing files in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-editor.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Create a file in the intended folder. 3. Enter or request the content. 4. Save it. 5. Run the project-specific verification command. 6. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/reviewing-changes title: Reviewing changes --- # Reviewing changes Open changed-file indicators or source control, compare additions and removals, check unexpected files, and request corrections before commit. ![Reviewing changes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-editor.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Open changed-file indicators or source control. 3. Compare additions and removals. 4. Check unexpected files, and request corrections before commit. 5. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/working-with-generated-files title: Working with generated files --- # Working with generated files Open the generated output, confirm format and destination, decide whether it belongs in source control, and regenerate rather than hand-edit when appropriate. ![Working with generated files in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-files.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Open the generated output. 3. Confirm format and destination, decide whether it belongs in source control, and regenerate rather than hand-edit when appropriate. 4. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/files-editor/opening-files-in-external-editors title: Opening files in external editors --- # Opening files in external editors Use Open externally or reveal the path, edit with the external application, save, then return to Foxora and confirm the file refreshes. ![Opening files in external editors in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-files.webp) ## Where this lives Open a project-backed task, then choose the project from the Stage. The file surface shows the active project tree, open tabs, file contents, agent edits, diagnostics, and generated project files. > **Safety:** Generated and agent-edited files are ordinary project files. Review diffs before overwriting, deleting, or committing them. ## Steps 1. Open a project-backed task, then choose the project from the Stage. 2. Use Open externally or reveal the path. 3. Edit with the external application. 4. Save. 5. Return to Foxora and confirm the file refreshes. 6. **Confirm the result.** Reload or reopen the file and run the appropriate formatter, test, build, or preview. ## Choose the right path - Edit directly for small precise changes. - Ask an agent for multi-file work. - Open an external editor when specialized tooling is needed. ## Confirm it worked - Reload or reopen the file and run the appropriate formatter, test, build, or preview. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/overview title: Overview --- # Overview Learn the Git panel workflow from status and diff review through staging, committing, branching, and safe recovery. ![Source Control in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. Learn the Git panel workflow from status and diff review through staging, committing, branching, and safe recovery. 3. Inspect the branch and clean/dirty state first, review every changed file, stage intentionally, and commit only a coherent verified change. 4. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/reviewing-git-changes-and-diffs title: Reviewing Git changes and diffs --- # Reviewing Git changes and diffs Open the source-control panel, inspect each changed and untracked file, compare line-level edits, and identify the task responsible. ![Reviewing Git changes and diffs in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. Open the source-control panel. 3. Inspect each changed and untracked file. 4. Compare line-level edits. 5. Identify the task responsible. 6. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/staging-and-committing-changes title: Staging and committing changes --- # Staging and committing changes Stage only related files or hunks, write an outcome-focused commit message, run verification, and inspect the created commit. ![Staging and committing changes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. Stage only related files or hunks. 3. Write an outcome-focused commit message. 4. Run verification. 5. Inspect the created commit. 6. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/branches-and-worktrees title: Branches and worktrees --- # Branches and worktrees Create or select a branch, use a separate worktree for parallel isolated work, confirm its path, and merge only after validation. ![Branches and worktrees in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. Create or select a branch. 3. Use a separate worktree for parallel isolated work. 4. Confirm its path, and merge only after validation. 5. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/resolving-conflicts title: Resolving conflicts --- # Resolving conflicts List conflicted files, understand both sides, edit the resolved result, remove conflict markers, test, stage, and continue the merge or rebase. ![Resolving conflicts in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. List conflicted files. 3. Understand both sides. 4. Edit the resolved result. 5. Remove conflict markers. 6. Test, stage, and continue the merge or rebase. 7. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/source-control/git-safety-and-recovery title: Git safety and recovery --- # Git safety and recovery Inspect reflog, status, branches, and uncommitted work before recovery; prefer a backup branch or stash over destructive reset. ![Git safety and recovery in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-source-control.webp) ## Where this lives Open a Git-backed project, then choose **Stage → Source control**. Source control shows repository status, diffs, staging, commits, branches, worktrees, and conflict state associated with the active project and task. > **Safety:** Never discard, reset, force-push, or overwrite changes merely to obtain a clean status. Preserve work before recovery operations. ## Steps 1. Open a Git-backed project, then choose **Stage → Source control**. 2. Inspect reflog, status, branches, and uncommitted work before recovery; prefer a backup branch or stash over destructive reset. 3. **Confirm the result.** Run tests, inspect `git status`, and review the final commit diff or pull request. ## Choose the right path - Use a branch for isolated work. - Use a worktree when parallel tasks need separate working directories. - Create a checkpoint before risky conflict resolution. ## Confirm it worked - Run tests, inspect `git status`, and review the final commit diff or pull request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/overview title: Overview --- # Overview Understand managed and existing-browser modes, Stage previews, takeover, authentication boundaries, and consequential web actions. ![Browser in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Understand managed and existing-browser modes, Stage previews, takeover, authentication boundaries, and consequential web actions. 3. Choose the browser mode, grant only required access, open the target page, and describe the exact outcome and actions that require confirmation. 4. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/browser-setup title: Browser setup --- # Browser setup Open Browser settings, select a supported mode, install or authorize the required bridge when prompted, and test with a harmless page. ![Browser setup in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Open Browser settings. 3. Select a supported mode. 4. Install or authorize the required bridge when prompted. 5. Test with a harmless page. 6. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/browser-operating-modes title: Browser operating modes --- # Browser operating modes Choose isolated managed browsing for clean state or an existing browser context for signed-in work, then verify which profile is active. ![Browser operating modes in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Choose isolated managed browsing for clean state or an existing browser context for signed-in work. 3. Verify which profile is active. 4. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/running-browser-tasks title: Running browser tasks --- # Running browser tasks Provide the target site and outcome, let Foxora navigate and inspect, follow actions on the Stage, and verify the final page state. ![Running browser tasks in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Provide the target site and outcome, let Foxora navigate and inspect. 3. Follow actions on the Stage. 4. Verify the final page state. 5. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/taking-over-an-authenticated-session title: Taking over an authenticated session --- # Taking over an authenticated session Pause agent control, take over the browser, complete login or sensitive input yourself, then hand control back only after checking the active account. ![Taking over an authenticated session in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Pause agent control. 3. Take over the browser. 4. Complete login or sensitive input yourself. 5. Hand control back only after checking the active account. 6. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/screenshots-and-visual-interaction title: Screenshots and visual interaction --- # Screenshots and visual interaction Capture the relevant viewport, point to or describe the target element, verify Foxora interpreted it correctly, and repeat after layout changes. ![Screenshots and visual interaction in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Capture the relevant viewport, point to or describe the target element. 3. Verify Foxora interpreted it correctly. 4. Repeat after layout changes. 5. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/browser/browser-permissions-and-safety title: Browser permissions and safety --- # Browser permissions and safety Review domain, account, scope, and proposed action before approval; keep purchases, publishing, messaging, and deletion manually gated. ![Browser permissions and safety in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/stage-browser.webp) ## Where this lives Open a task, then choose **Stage → Browser**. Foxora can operate an isolated managed browser or an authorized existing browser context, while showing navigation, screenshots, actions, and takeover state on the Stage. > **Safety:** Browser actions can create irreversible external changes. Never expose more browser state or account access than the task requires. ## Steps 1. Open a task, then choose **Stage → Browser**. 2. Review domain, account, scope, and proposed action before approval. 3. Keep purchases, publishing, messaging, and deletion manually gated. 4. **Confirm the result.** Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. ## Choose the right path - Use managed mode for isolated research and testing. - Use an existing signed-in context only when account state is necessary. - Take over for authentication, captchas, or sensitive confirmation. ## Confirm it worked - Inspect the final URL and page state, then independently confirm any submitted, published, purchased, or deleted result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/overview title: Overview --- # Overview Understand observation, pointer and keyboard control, OS permission prompts, takeover, and the limits of desktop automation. ![Computer Use in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Understand observation, pointer and keyboard control, OS permission prompts, takeover, and the limits of desktop automation. 3. Grant screen and control permissions only when needed, place the target application in a known state, and specify the exact task boundary. 4. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/controlling-desktop-applications title: Controlling desktop applications --- # Controlling desktop applications Open the target application, arrange the relevant window, state the exact result, monitor each transition, and verify the application’s saved state. ![Controlling desktop applications in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Open the target application. 3. Arrange the relevant window, state the exact result. 4. Monitor each transition. 5. Verify the application’s saved state. 6. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/keyboard-and-pointer-interaction title: Keyboard and pointer interaction --- # Keyboard and pointer interaction Provide a stable visual target, let Foxora click, type, scroll, or send keys, and stop immediately if focus moves to the wrong control. ![Keyboard and pointer interaction in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Provide a stable visual target, let Foxora click, type, scroll, or send keys. 3. Stop immediately if focus moves to the wrong control. 4. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/permission-requests title: Permission requests --- # Permission requests Read the genuine operating-system permission prompt, confirm the requested capability and Foxora identity, grant the minimum access, and retest. ![Permission requests in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Read the genuine operating-system permission prompt. 3. Confirm the requested capability and Foxora identity. 4. Grant the minimum access, and retest. 5. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/taking-over-from-foxora title: Taking over from Foxora --- # Taking over from Foxora Interrupt the task, take control at login or sensitive steps, complete the action, place the app in a clear state, and explicitly return control. ![Taking over from Foxora in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Interrupt the task. 3. Take control at login or sensitive steps. 4. Complete the action, place the app in a clear state, and explicitly return control. 5. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/computer-use/computer-control-safety title: Computer-control safety --- # Computer-control safety Close unrelated sensitive windows, use a test account where possible, keep approvals enabled, monitor the screen, and verify every external side effect. ![Computer-control safety in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings → Permissions** before starting a computer-control task. Computer Use lets Foxora observe the screen and operate pointer, keyboard, scrolling, and supported desktop applications under operating-system permissions. > **Safety:** Keep financial, administrative, credential, destructive, and personally sensitive actions under direct human control. ## Steps 1. Open **Settings → Permissions** before starting a computer-control task. 2. Close unrelated sensitive windows. 3. Use a test account where possible. 4. Keep approvals enabled. 5. Monitor the screen. 6. Verify every external side effect. 7. **Confirm the result.** Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. ## Choose the right path - Use screenshots for observation-only work. - Allow pointer and keyboard control for bounded interaction. - Take over whenever authentication or ambiguity appears. ## Confirm it worked - Watch the Stage or screen, inspect the destination application, and confirm saved or submitted state directly. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/overview title: Overview --- # Overview Build reliable specialists by separating an agent’s role, instructions, model, knowledge, skills, and tool permissions. ![Agents in Foxora v5.0.3](/docs/images/v5.0.3/studio/agents-overview.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Build reliable specialists by separating an agent’s role, instructions, model, knowledge, skills, and tool permissions. 3. Choose **New agent**, then **Build with AI** or **Start from scratch**. In the editor, complete Persona, Instructions, Model, Tools, Skills, Plugins, and Test; assign one core role and keep capabilities narrow. 4. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/creating-an-agent-with-ai title: Creating an agent with AI --- # Creating an agent with AI Describe the specialist, its inputs, expected outputs, and boundaries; generate the draft, inspect every field, and save only after correcting assumptions. ![Creating an agent with AI in Foxora v5.0.3](/docs/images/v5.0.3/studio/new-agent.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Describe the specialist, its inputs, expected outputs, and boundaries; generate the draft. 3. Inspect every field. 4. Save only after correcting assumptions. 5. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/creating-an-agent-manually title: Creating an agent manually --- # Creating an agent manually Create a blank agent, name its single responsibility, write instructions, select a model, add capabilities, and save a version you can test. ![Creating an agent manually in Foxora v5.0.3](/docs/images/v5.0.3/studio/new-agent.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Create a blank agent. 3. Name its single responsibility. 4. Write instructions. 5. Select a model. 6. Add capabilities. 7. Save a version you can test. 8. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/agent-instructions-and-configuration title: Agent instructions and configuration --- # Agent instructions and configuration Write the role, procedure, constraints, output format, and escalation rules; keep durable behavior in instructions and task-specific details in the task prompt. ![Agent instructions and configuration in Foxora v5.0.3](/docs/images/v5.0.3/studio/agent-builder.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Write the role, procedure, constraints, output format, and escalation rules. 3. Keep durable behavior in instructions and task-specific details in the task prompt. 4. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/specialist-roles-and-delegation title: Specialist roles and delegation --- # Specialist roles and delegation Give each specialist a non-overlapping responsibility, define when it should receive work, and make the expected handoff explicit for crew coordination. ![Specialist roles and delegation in Foxora v5.0.3](/docs/images/v5.0.3/studio/agent-builder.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Give each specialist a non-overlapping responsibility. 3. Define when it should receive work, and make the expected handoff explicit for crew coordination. 4. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/testing-an-agent title: Testing an agent --- # Testing an agent Prepare a normal case, an ambiguous case, and a permission-sensitive case; run each in a clean task and compare the result with explicit acceptance criteria. ![Testing an agent in Foxora v5.0.3](/docs/images/v5.0.3/studio/agent-editor.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Prepare a normal case, an ambiguous case, and a permission-sensitive case. 3. Run each in a clean task and compare the result with explicit acceptance criteria. 4. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/editing-and-managing-agents title: Editing and managing agents --- # Editing and managing agents Duplicate before major experiments, edit one behavior at a time, retest dependent crews or bots, and disable or remove definitions that should no longer be selected. ![Editing and managing agents in Foxora v5.0.3](/docs/images/v5.0.3/studio/agents-overview.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Duplicate before major experiments. 3. Edit one behavior at a time, retest dependent crews or bots. 4. Disable or remove definitions that should no longer be selected. 5. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/agents/agent-models-and-tiers title: Agent models and tiers --- # Agent models and tiers Choose a managed Foxora processing tier or assign a BYOK provider and model to this agent; test quality, latency, and credit or provider usage before wider use. ![Agent models and tiers in Foxora v5.0.3](/docs/images/v5.0.3/studio/agent-editor.webp) ## Where this lives Open **studio → Agents**. An agent is a reusable specialist with its own name, role, instructions, model, skills, tools, and optional provider key. Agents can work directly in tasks or as members of a crew. > **Safety:** Do not place secrets in agent instructions. Add provider keys through the agent model settings and keep tool access limited to the agent’s actual job. ## Steps 1. Open **studio → Agents**. 2. Choose a managed Foxora processing tier or assign a BYOK provider and model to this agent. 3. Test quality, latency, and credit or provider usage before wider use. 4. **Confirm the result.** Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. ## Choose the right path - Create with AI when you can describe the outcome but not every field. - Create manually when exact instructions and permissions matter. - Use a crew when the objective needs several distinct specialists. ## Confirm it worked - Run the same test in a fresh task and confirm the agent follows its role, uses the intended tools, requests the expected approvals, and produces a verifiable result. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/overview title: Overview --- # Overview Understand how a lead, specialists, delegation, processing tier, and final synthesis work together in a crew. ![Crews in Foxora v5.0.3](/docs/images/v5.0.3/studio/crews-overview.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Understand how a lead, specialists, delegation, processing tier, and final synthesis work together in a crew. 3. Choose **New crew**, then **Build with AI** or **Start from scratch**. Set the name, mission, goal, Vixen/Zenko/Kyubi tier, composer visibility, lead, and team; use the org chart to check the resulting organization. 4. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/how-crews-coordinate-work title: How crews coordinate work --- # How crews coordinate work Trace how the lead interprets a request, assigns bounded subtasks, receives specialist results, resolves conflicts, and produces the final response. ![How crews coordinate work in Foxora v5.0.3](/docs/images/v5.0.3/studio/crews-overview.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Trace how the lead interprets a request, assigns bounded subtasks, receives specialist results, resolves conflicts, and produces the final response. 3. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/creating-a-crew-with-ai title: Creating a crew with AI --- # Creating a crew with AI Describe the recurring outcome and required expertise, generate a proposed team, inspect every role and permission, then test the draft before regular use. ![Creating a crew with AI in Foxora v5.0.3](/docs/images/v5.0.3/studio/new-crew.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Describe the recurring outcome and required expertise, generate a proposed team. 3. Inspect every role and permission. 4. Test the draft before regular use. 5. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/creating-a-crew-manually title: Creating a crew manually --- # Creating a crew manually Create a crew, select the lead, add only necessary specialists, define coordination instructions, choose a tier, and save the initial configuration. ![Creating a crew manually in Foxora v5.0.3](/docs/images/v5.0.3/studio/new-crew.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Create a crew. 3. Select the lead. 4. Add only necessary specialists. 5. Define coordination instructions. 6. Choose a tier. 7. Save the initial configuration. 8. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/adding-and-configuring-agents title: Adding and configuring agents --- # Adding and configuring agents Add existing agents or create specialists, remove overlapping roles, order or describe handoffs, and confirm each member has only the capabilities it needs. ![Adding and configuring agents in Foxora v5.0.3](/docs/images/v5.0.3/studio/crew-builder.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Add existing agents or create specialists. 3. Remove overlapping roles, order or describe handoffs. 4. Confirm each member has only the capabilities it needs. 5. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/choosing-a-crew-in-a-session title: Choosing a crew for a task --- # Choosing a crew for a task Open the composer crew picker, select the crew that matches the task, confirm its tier and lead, then send an objective with a measurable definition of done. ![Choosing a crew for a task in Foxora v5.0.3](/docs/images/v5.0.3/workspace/crew-picker.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Open the composer crew picker. 3. Select the crew that matches the task. 4. Confirm its tier and lead. 5. Send an objective with a measurable definition of done. 6. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/crews/testing-and-managing-crews title: Testing and managing crews --- # Testing and managing crews Use a representative multi-part task, inspect delegation and synthesis, adjust one role or rule at a time, and retest before duplicating, disabling, or deleting the crew. ![Testing and managing crews in Foxora v5.0.3](/docs/images/v5.0.3/studio/crew-builder.webp) ## Where this lives Open **studio → Crews**. A crew coordinates a lead agent and one or more specialists. The lead decomposes the objective, delegates bounded work, combines results, and remains responsible for the final answer. > **Safety:** Every member’s tools and external access still matter. A lead should not delegate sensitive actions to a broadly privileged specialist without an approval boundary. ## Steps 1. Open **studio → Crews**. 2. Use a representative multi-part task. 3. Inspect delegation and synthesis. 4. Adjust one role or rule at a time, and retest before duplicating, disabling, or deleting the crew. 5. **Confirm the result.** Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. ## Choose the right path - Use one agent for a narrow task. - Use a crew when parallel expertise or review materially improves the result. - Create separate crews for recurring workflows with different risk or quality requirements. ## Confirm it worked - Start a fresh task with the crew, follow each delegated branch, and verify that the lead integrates evidence instead of merely concatenating specialist replies. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/overview title: Overview --- # Overview Configure a reusable bot, connect it to supported surfaces, test its behavior, and operate it with clear permissions and escalation rules. ![Bots in Foxora v5.0.3](/docs/images/v5.0.3/studio/bots-overview.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Configure a reusable bot. 3. Connect it to supported surfaces. 4. Test its behavior, and operate it with clear permissions and escalation rules. 5. Choose **New bot**, select a crew, define the purpose and instructions, choose a branding template, colors, audience, and deployment surfaces, then set an exact HTTPS website origin for web deployments. 6. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/how-bots-differ-from-agents title: How bots differ from agents --- # How bots differ from agents Compare private agent tasks with published bot interactions, then choose the surface whose audience, lifecycle, and security model fit the use case. ![How bots differ from agents in Foxora v5.0.3](/docs/images/v5.0.3/studio/bots-overview.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Compare private agent tasks with published bot interactions. 3. Choose the surface whose audience, lifecycle, and security model fit the use case. 4. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/creating-a-bot title: Creating a bot --- # Creating a bot Create the bot, assign its agent or crew, choose the supported surface, set its name and audience, and save an unpublished draft for testing. ![Creating a bot in Foxora v5.0.3](/docs/images/v5.0.3/studio/bot-builder.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Create the bot. 3. Assign its agent or crew. 4. Choose the supported surface. 5. Set its name and audience. 6. Save an unpublished draft for testing. 7. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/configuring-bot-behavior title: Configuring bot behavior --- # Configuring bot behavior Define supported requests, response style, unavailable actions, escalation rules, and the exact information the bot may expose. ![Configuring bot behavior in Foxora v5.0.3](/docs/images/v5.0.3/studio/bot-builder.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Define supported requests, response style, unavailable actions, escalation rules, and the exact information the bot may expose. 3. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/connecting-tools-and-channels title: Connecting tools and channels --- # Connecting tools and channels Add only the plugin or native tools the bot needs, connect a channel, map the destination, and confirm the bot identity and permissions. ![Connecting tools and channels in Foxora v5.0.3](/docs/images/v5.0.3/studio/bot-builder.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Add only the plugin or native tools the bot needs. 3. Connect a channel. 4. Map the destination. 5. Confirm the bot identity and permissions. 6. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/testing-a-bot title: Testing a bot --- # Testing a bot Test normal, unsupported, ambiguous, and permission-sensitive messages in a safe destination; verify replies, tool calls, routing, and failure handling. ![Testing a bot in Foxora v5.0.3](/docs/images/v5.0.3/studio/bot-builder.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Test normal, unsupported, ambiguous, and permission-sensitive messages in a safe destination. 3. Verify replies, tool calls, routing, and failure handling. 4. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/bots/managing-and-disabling-bots title: Managing and disabling bots --- # Managing and disabling bots Inspect active connections and recent behavior, update one configuration at a time, disable before maintenance, and verify traffic no longer routes to the bot. ![Managing and disabling bots in Foxora v5.0.3](/docs/images/v5.0.3/studio/bots-overview.webp) ## Where this lives Open **studio → Bots**. A bot is a published conversational surface that applies configured agent behavior to connected channels or embedded experiences. It is designed for repeated inbound interactions rather than a private builder task. > **Safety:** Treat every channel message as untrusted input. Keep high-impact tools gated, avoid exposing internal instructions, and test authorization boundaries between users. ## Steps 1. Open **studio → Bots**. 2. Inspect active connections and recent behavior. 3. Update one configuration at a time. 4. Disable before maintenance. 5. Verify traffic no longer routes to the bot. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. ## Choose the right path - Use an agent for private interactive work. - Use a bot for repeatable conversations through a connected surface. - Use a workflow for deterministic event-driven sequences. ## Confirm it worked - Send test messages from every connected surface, confirm routing and account context, inspect tool activity, and verify the visible reply and any external action. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/overview title: Overview --- # Overview Understand how skills add repeatable procedures and supporting resources to agents without changing the underlying agent identity. ![Skills in Foxora v5.0.3](/docs/images/v5.0.3/studio/skills-overview.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Understand how skills add repeatable procedures and supporting resources to agents without changing the underlying agent identity. 3. Choose **New skill**, then **Build with AI** or **Start from scratch**. Set identity, version, tags, palace locus, tools, scripts, and the SKILL.md instructions that an agent will load on demand. 4. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/built-in-skills title: Built-in skills --- # Built-in skills Browse the built-in catalog, open the skill details, compare the supported use case and requirements, attach it to a test agent, and run its example workflow. ![Built-in skills in Foxora v5.0.3](/docs/images/v5.0.3/studio/skills-overview.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Browse the built-in catalog. 3. Open the skill details. 4. Compare the supported use case and requirements. 5. Attach it to a test agent. 6. Run its example workflow. 7. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/installing-a-skill title: Installing a skill --- # Installing a skill Select a trusted source, review instructions and bundled files, inspect required tools and permissions, install it, and verify the installed version. ![Installing a skill in Foxora v5.0.3](/docs/images/v5.0.3/studio/skills-overview.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Select a trusted source. 3. Review instructions and bundled files. 4. Inspect required tools and permissions. 5. Install it. 6. Verify the installed version. 7. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/creating-a-custom-skill title: Creating a custom skill --- # Creating a custom skill Define a narrow trigger and outcome, write an explicit procedure, add only necessary resources or scripts, validate references, and test with a dedicated agent. ![Creating a custom skill in Foxora v5.0.3](/docs/images/v5.0.3/studio/new-skill.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Define a narrow trigger and outcome. 3. Write an explicit procedure. 4. Add only necessary resources or scripts. 5. Validate references. 6. Test with a dedicated agent. 7. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/skill-instructions-and-resources title: Skill instructions and resources --- # Skill instructions and resources Separate required procedure from background reference material, keep paths and commands portable, document inputs and outputs, and verify every linked resource exists. ![Skill instructions and resources in Foxora v5.0.3](/docs/images/v5.0.3/studio/skill-builder.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Separate required procedure from background reference material. 3. Keep paths and commands portable. 4. Document inputs and outputs. 5. Verify every linked resource exists. 6. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/testing-a-skill title: Testing a skill --- # Testing a skill Run a standard case, an edge case, and a failure case; confirm the skill activates at the right time and does not overreach beyond its stated workflow. ![Testing a skill in Foxora v5.0.3](/docs/images/v5.0.3/studio/skill-detail.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Run a standard case, an edge case, and a failure case. 3. Confirm the skill activates at the right time and does not overreach beyond its stated workflow. 4. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/skills/managing-and-removing-skills title: Managing and removing skills --- # Managing and removing skills Review attached agents before updating, test a new version in isolation, detach an obsolete skill, and remove it only after dependent configurations are migrated. ![Managing and removing skills in Foxora v5.0.3](/docs/images/v5.0.3/studio/skills-overview.webp) ## Where this lives Open **studio → Skills**. A skill is a packaged set of instructions and optional resources, scripts, or tool guidance that teaches agents a repeatable way to perform a specialized workflow. > **Safety:** Treat installed skill instructions and scripts as executable supply-chain input. Review their source, requested tools, paths, and network behavior before enabling them. ## Steps 1. Open **studio → Skills**. 2. Review attached agents before updating. 3. Test a new version in isolation, detach an obsolete skill. 4. Remove it only after dependent configurations are migrated. 5. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 6. **Confirm the result.** Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. ## Choose the right path - Use a built-in skill for a supported standard workflow. - Install a reviewed skill for reusable third-party guidance. - Create a custom skill when your organization has a stable process worth encoding. ## Confirm it worked - Invoke the skill from a fresh agent task, confirm the agent follows its procedure, and verify the output and any tool action independently. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/overview title: Overview --- # Overview Understand tool discovery, typed inputs and results, approval boundaries, failure reporting, and the difference between built-in, plugin, and MCP tools. ![Tools in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Understand tool discovery, typed inputs and results, approval boundaries, failure reporting, and the difference between built-in, plugin, and MCP tools. 3. Search the built-in tool catalog by capability, inspect the tool’s expected inputs and permission boundary, then enable it only through the agent, skill, or workflow that needs it. 4. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/built-in-tool-catalog title: Built-in tool catalog --- # Built-in tool catalog Browse tools by capability, open the input and permission details, identify read-only versus mutating actions, and attach only the required tools to an agent. ![Built-in tool catalog in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Browse tools by capability. 3. Open the input and permission details. 4. Identify read-only versus mutating actions. 5. Attach only the required tools to an agent. 6. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/how-agents-select-tools title: How agents select tools --- # How agents select tools Give the agent a clear outcome, confirm relevant tools are enabled, inspect the chosen call and arguments, and refine instructions if it repeatedly chooses the wrong capability. ![How agents select tools in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Give the agent a clear outcome. 3. Confirm relevant tools are enabled. 4. Inspect the chosen call and arguments, and refine instructions if it repeatedly chooses the wrong capability. 5. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/tool-inputs-and-results title: Tool inputs and results --- # Tool inputs and results Review each required input and target, run the call, expand the result shape, and use status, identifiers, paths, or URLs to verify the operation. ![Tool inputs and results in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Review each required input and target. 3. Run the call. 4. Expand the result shape. 5. Use status, identifiers, paths, or URLs to verify the operation. 6. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/tool-permissions-and-approvals title: Tool permissions and approvals --- # Tool permissions and approvals Match the tool to the task autonomy mode, read the exact target and proposed action, approve only understood requests, and revoke unused access afterward. ![Tool permissions and approvals in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Match the tool to the task autonomy mode. 3. Read the exact target and proposed action. 4. Approve only understood requests. 5. Revoke unused access afterward. 6. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/handling-tool-failures title: Handling tool failures --- # Handling tool failures Capture the tool name, inputs, error code, and timestamp; check permissions and connectivity, correct one cause, then repeat the smallest failing call. ![Handling tool failures in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Capture the tool name, inputs, error code, and timestamp. 4. Check permissions and connectivity. 5. Correct one cause. 6. Repeat the smallest failing call. 7. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 8. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/tools/creating-or-connecting-custom-tools title: Creating or connecting custom tools --- # Creating or connecting custom tools Choose MCP for a custom server or Composio for a supported integration, define a narrow schema and permissions, connect it, and test with a non-destructive operation. ![Creating or connecting custom tools in Foxora v5.0.3](/docs/images/v5.0.3/studio/tools-overview.webp) ## Where this lives Open **studio → Tools**. Tools are typed capabilities that let an agent inspect or change files, run commands, browse, control applications, query services, or call connected integrations. > **Safety:** Tool output is untrusted data and can contain misleading instructions. Do not grant a tool broader filesystem, account, or network access merely to avoid configuration work. ## Steps 1. Open **studio → Tools**. 2. Choose MCP for a custom server or Composio for a supported integration. 3. Define a narrow schema and permissions. 4. Connect it. 5. Test with a non-destructive operation. 6. **Confirm the result.** Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. ## Choose the right path - Use built-in tools for local and first-party capabilities. - Use Composio-managed plugins for supported SaaS integrations. - Use MCP for a custom or organization-provided tool server. ## Confirm it worked - Run one representative call, inspect the structured result or error, and confirm the underlying file, process, page, application, or external service changed as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/overview title: Overview --- # Overview Understand local memory capture, recall, provenance, review, correction, deletion, and the boundary between memory and task history. ![Memory Den in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Understand local memory capture, recall, provenance. 3. Review, correction, deletion, and the boundary between memory and task history. 4. Use the palace loci to browse durable knowledge. Choose **Place memory** to set a locus, anchor, content, tags, durability, memory layer, and confidence; do not store credentials or unrelated personal data. 5. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/how-automatic-memory-works title: How automatic memory works --- # How automatic memory works Complete a focused interaction, inspect what Foxora selected as durable, distinguish inferred context from explicit facts, and correct anything that should not persist. ![How automatic memory works in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Complete a focused interaction. 3. Inspect what Foxora selected as durable. 4. Distinguish inferred context from explicit facts. 5. Correct anything that should not persist. 6. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/memory-recall-in-sessions title: Memory recall in tasks --- # Memory recall in tasks Open a later task on the same device, ask a relevant question, inspect the recalled memory card or activity, and provide updated information when the fact has changed. ![Memory recall in tasks in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Open a later task on the same device, ask a relevant question. 3. Inspect the recalled memory card or activity. 4. Provide updated information when the fact has changed. 5. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/exploring-saved-memories title: Exploring saved memories --- # Exploring saved memories Open Memory Den, search or filter entries, select a memory, read its content and metadata, and follow the source reference before relying on it. ![Exploring saved memories in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Open Memory Den. 3. Search or filter entries. 4. Select a memory. 5. Read its content and metadata. 6. Follow the source reference before relying on it. 7. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/memory-sources-and-provenance title: Memory sources and provenance --- # Memory sources and provenance Open an entry, inspect its source task and creation context, compare it with the original transcript, and mark unsupported or outdated conclusions for correction. ![Memory sources and provenance in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Open an entry. 3. Inspect its source task and creation context. 4. Compare it with the original transcript, and mark unsupported or outdated conclusions for correction. 5. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/editing-and-deleting-memories title: Editing and deleting memories --- # Editing and deleting memories Select the exact local memory, edit the smallest inaccurate portion or delete it, confirm the action, then test recall in a new local task. ![Editing and deleting memories in Foxora v5.0.3](/docs/images/v5.0.3/studio/place-memory.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Select the exact local memory. 3. Edit the smallest inaccurate portion or delete it. 4. Confirm the action. 5. Test recall in a new local task. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/memory-den/memory-privacy-and-storage title: Memory privacy and storage --- # Memory privacy and storage Review what is stored locally, restrict device access and backups, export only when necessary, and clear local memory separately from account or subscription changes. ![Memory privacy and storage in Foxora v5.0.3](/docs/images/v5.0.3/studio/memory-overview.webp) ## Where this lives Open **studio → Memory**. The Memory Den stores durable memories locally on this device. Foxora can recall relevant items in later local tasks and preserves source information so you can inspect why a memory exists. > **Safety:** Tasks and Memory Den data are local only. Desktop and CLI share them through the same local runtime on a device; they do not automatically sync to the web app or another machine. ## Steps 1. Open **studio → Memory**. 2. Review what is stored locally. 3. Restrict device access and backups. 4. Export only when necessary. 5. Clear local memory separately from account or subscription changes. 6. **Confirm the result.** Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. ## Choose the right path - Use task history for task-specific detail. - Use Memory Den for durable preferences and facts on this device. - Use project instructions for explicit repository rules that should not depend on recall. ## What stays local - Session transcripts and checkpoints. - Memory Den entries and their local provenance. - Local attachments and runtime state associated with the device. Desktop and CLI can both access this data when they use the same local runtime. Signing into the same account on the web app or another computer does not copy it there automatically. ## Confirm it worked - Start a new task from Desktop or CLI on the same local runtime, ask a question that should use the memory, inspect the recalled item and provenance, and confirm another device does not receive it automatically. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/overview title: Overview --- # Overview Create dependable scheduled automation by defining the task, target, timezone, recurrence, permissions, delivery, and run monitoring. ![Scheduled Tasks in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedules-overview.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Create dependable scheduled automation by defining the task, target, timezone, recurrence, permissions, delivery. 3. Run monitoring. 4. Choose **New scheduled task**, set title, description, agent or crew assignee, state, priority, and free-text timing, then review the queued item before relying on unattended execution. 5. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/creating-a-scheduled-task title: Creating a scheduled task --- # Creating a scheduled task Create the task, name the outcome, write the input, select an agent or crew, choose the first run time and destination, then save it disabled until tested. ![Creating a scheduled task in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedule-editor.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Create the task. 3. Name the outcome. 4. Write the input. 5. Select an agent or crew. 6. Choose the first run time and destination. 7. Save it disabled until tested. 8. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/schedules-and-recurrence title: Schedules and recurrence --- # Schedules and recurrence Choose one-time or recurring execution, set the timezone, configure frequency and boundaries, preview upcoming times, and check daylight-saving behavior where relevant. ![Schedules and recurrence in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedule-editor.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Choose one-time or recurring execution. 3. Set the timezone, configure frequency and boundaries, preview upcoming times. 4. Check daylight-saving behavior where relevant. 5. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/selecting-an-agent-or-crew title: Selecting an agent or crew --- # Selecting an agent or crew Choose a tested configuration whose tools and model fit the task, confirm external connections, and avoid changing that configuration without retesting the schedule. ![Selecting an agent or crew in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedule-editor.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Choose a tested configuration whose tools and model fit the task. 3. Confirm external connections, and avoid changing that configuration without retesting the schedule. 4. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/editing-pausing-and-deleting-tasks title: Editing, pausing, and deleting tasks --- # Editing, pausing, and deleting tasks Pause before major edits, change one schedule or input field, run a manual test, resume after verification, and delete only after preserving required run outputs. ![Editing, pausing, and deleting tasks in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedules-overview.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Pause before major edits. 3. Change one schedule or input field. 4. Run a manual test, resume after verification. 5. Delete only after preserving required run outputs. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/scheduled-task-permissions title: Scheduled-task permissions --- # Scheduled-task permissions Review every tool and external destination, keep high-impact actions gated or narrowly scoped, verify connected-account access, and use the least powerful agent configuration. ![Scheduled-task permissions in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedule-editor.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Review every tool and external destination. 3. Keep high-impact actions gated or narrowly scoped. 4. Verify connected-account access. 5. Use the least powerful agent configuration. 6. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/scheduled-tasks/troubleshooting-scheduled-tasks title: Troubleshooting scheduled tasks --- # Troubleshooting scheduled tasks Inspect next-run time, timezone, enabled state, selected agent or crew, connection status, and the latest run error; fix one cause and use Run now to retest. ![Troubleshooting scheduled tasks in Foxora v5.0.3](/docs/images/v5.0.3/automation/schedules-overview.webp) ## Where this lives Open **foxora → Schedules**. A scheduled task starts a configured agent or crew automatically at a specified time or recurrence and records each execution as a run with status, logs, and output. > **Safety:** Scheduled work may run without a person watching. Avoid unrestricted destructive actions, confirm connection availability, and make repeated execution safe. ## Steps 1. Open **foxora → Schedules**. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Inspect next-run time, timezone, enabled state, selected agent or crew, connection status, and the latest run error; fix one cause and use Run now to retest. 4. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 5. **Confirm the result.** Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. ## Choose the right path - Use a one-time schedule for deferred work. - Use recurrence for predictable maintenance or reporting. - Use an event-triggered workflow when timing is not the real trigger. ## Confirm it worked - Use Run now for a controlled test, inspect the created run and destination, then confirm the next scheduled time in the intended timezone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/overview title: Overview --- # Overview Design, test, publish, and monitor workflows made from triggers, steps, conditions, delays, approvals, and connected actions. ![Workflows in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflows-overview.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Design. 3. Test. 4. Publish. 5. Monitor workflows made from triggers, steps, conditions, delays, approvals, and connected actions. 6. Choose **New workflow**, then **Build with AI** or **Start from scratch**. Configure the Trigger and Agent step on the canvas, add nodes, inspect Workflow or JSON, save a draft, run a test, and publish only the reviewed version. 7. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/creating-a-workflow-with-ai title: Creating a workflow with AI --- # Creating a workflow with AI Describe the trigger, required inputs, ordered result, error behavior, and approval points; generate the draft, then inspect every node and mapping before testing. ![Creating a workflow with AI in Foxora v5.0.3](/docs/images/v5.0.3/automation/new-workflow.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Describe the trigger, required inputs, ordered result, error behavior, and approval points; generate the draft. 3. Inspect every node and mapping before testing. 4. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/creating-a-workflow-manually title: Creating a workflow manually --- # Creating a workflow manually Create a blank workflow, add and configure the trigger, build one step at a time, map typed inputs, define outputs, and save an unpublished draft. ![Creating a workflow manually in Foxora v5.0.3](/docs/images/v5.0.3/automation/new-workflow.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Create a blank workflow. 3. Add and configure the trigger. 4. Build one step at a time. 5. Map typed inputs. 6. Define outputs. 7. Save an unpublished draft. 8. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/triggers-steps-and-actions title: Triggers, steps, and actions --- # Triggers, steps, and actions Choose the initiating event, inspect its payload, connect each step’s required input to a validated value, and make the final side effect explicit. ![Triggers, steps, and actions in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflow-editor.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Choose the initiating event. 3. Inspect its payload. 4. Connect each step’s required input to a validated value, and make the final side effect explicit. 5. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/delays-and-conditional-behavior title: Delays and conditional behavior --- # Delays and conditional behavior Add a condition with complete true and false paths, configure delays with a timezone or duration, and define what happens when data is missing or late. ![Delays and conditional behavior in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflow-editor.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Add a condition with complete true and false paths, configure delays with a timezone or duration. 3. Define what happens when data is missing or late. 4. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/approvals-inside-workflows title: Approvals inside workflows --- # Approvals inside workflows Insert approval immediately before the consequential step, show the exact proposed action and target, define expiry or rejection behavior, and test both decisions. ![Approvals inside workflows in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflow-editor.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Insert approval immediately before the consequential step, show the exact proposed action and target. 3. Define expiry or rejection behavior. 4. Test both decisions. 5. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/testing-a-workflow title: Testing a workflow --- # Testing a workflow Use representative test data, run each branch including failures and retries, inspect step-level logs, and verify no real external side effect occurs unexpectedly. ![Testing a workflow in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflow-editor.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Use representative test data. 3. Run each branch including failures and retries. 4. Inspect step-level logs. 5. Verify no real external side effect occurs unexpectedly. 6. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/workflows/publishing-and-managing-workflows title: Publishing and managing workflows --- # Publishing and managing workflows Name and version the tested draft, publish it, monitor initial runs, pause before structural edits, and retire versions only after dependent triggers are migrated. ![Publishing and managing workflows in Foxora v5.0.3](/docs/images/v5.0.3/automation/workflows-overview.webp) ## Where this lives Open **foxora → Workflows**. A workflow is a reusable automation made of a trigger and ordered steps such as agent work, tool actions, conditions, delays, approvals, and delivery. > **Safety:** Assume triggers and external data are untrusted. Validate inputs, limit credentials, prevent duplicate side effects, and require approval before irreversible operations. ## Steps 1. Open **foxora → Workflows**. 2. Name and version the tested draft. 3. Publish it. 4. Monitor initial runs. 5. Pause before structural edits. 6. Retire versions only after dependent triggers are migrated. 7. **Confirm the result.** Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. ## Choose the right path - Use a task for exploratory work. - Use a scheduled task for one prompt on a clock. - Use a workflow for repeatable multi-step logic or event-driven automation. ## Confirm it worked - Run the workflow with test data, inspect every step input and output, verify the final destination, and publish only the tested version. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/overview title: Overview --- # Overview Use run records to start work, monitor progress, inspect outputs and usage, diagnose failures, retry safely, or cancel execution. ![Runs in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Use run records to start work. 3. Monitor progress. 4. Inspect outputs and usage, diagnose failures, retry safely, or cancel execution. 5. Open **Runs** from the task header to inspect background tasks. Expand the panel when needed, follow active work, and close the panel without assuming that closing it cancels the underlying run. 6. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/starting-a-run title: Starting a run --- # Starting a run Choose the tested agent, crew, workflow, or task, enter validated input, confirm model and destination, start once, and open the created run record. ![Starting a run in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Choose the tested agent, crew, workflow, or task. 3. Enter validated input. 4. Confirm model and destination. 5. Start once. 6. Open the created run record. 7. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/monitoring-live-progress title: Monitoring live progress --- # Monitoring live progress Follow the current step and elapsed time, expand agent and tool activity, answer approval requests, and interrupt only after checking whether work is still advancing. ![Monitoring live progress in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Follow the current step and elapsed time. 3. Expand agent and tool activity, answer approval requests. 4. Interrupt only after checking whether work is still advancing. 5. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/run-history-and-outputs title: Run history and outputs --- # Run history and outputs Filter runs by source, status, and date, open the required record, inspect its input and output, then follow identifiers or links to the persistent destination. ![Run history and outputs in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Filter runs by source, status, and date. 3. Open the required record. 4. Inspect its input and output. 5. Follow identifiers or links to the persistent destination. 6. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/reviewing-failed-runs title: Reviewing failed runs --- # Reviewing failed runs Open the first failed step, capture its error and preceding input, distinguish configuration from transient failure, and preserve logs before changing anything. ![Reviewing failed runs in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/task-transcript.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Open the first failed step. 4. Capture its error and preceding input. 5. Distinguish configuration from transient failure. 6. Preserve logs before changing anything. 7. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 8. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/retrying-or-cancelling-a-run title: Retrying or cancelling a run --- # Retrying or cancelling a run Check whether completed steps are safe to repeat, correct the input or configuration, retry once from the supported boundary, or cancel and confirm execution stops. ![Retrying or cancelling a run in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Check whether completed steps are safe to repeat. 3. Correct the input or configuration, retry once from the supported boundary, or cancel and confirm execution stops. 4. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 5. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/runs/run-logs-and-diagnostics title: Run logs and diagnostics --- # Run logs and diagnostics Open the run timeline and logs, correlate timestamps across agents and tools, copy a redacted diagnostic bundle, and include the run identifier in support requests. ![Run logs and diagnostics in Foxora v5.0.3](/docs/images/v5.0.3/operations/runs-panel.webp) ## Where this lives Open a task and choose **Runs** in its header. A run is one execution record for an agent, crew, scheduled task, workflow, bot, or channel-triggered job. It contains status, timing, step activity, logs, usage, and outputs. > **Safety:** Retrying can repeat external actions. Inspect the last confirmed successful step and use idempotency or a corrected input before running again. ## Steps 1. Open a task and choose **Runs** in its header. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Open the run timeline and logs, correlate timestamps across agents and tools. 4. Copy a redacted diagnostic bundle, and include the run identifier in support requests. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. ## Choose the right path - Start a new run for a new input. - Retry only a failed or safely repeatable execution. - Cancel when continuing could waste credits or create an unwanted side effect. ## Confirm it worked - Check terminal status, output identifiers, logs, usage, and the actual destination rather than relying on a green badge alone. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/overview title: Overview --- # Overview Connect supported applications through Composio, expose selected actions as tools, and manage account access without putting credentials in prompts. ![Plugins in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Connect supported applications through Composio, expose selected actions as tools, and manage account access without putting credentials in prompts. 3. Open the catalog, filter more than 1,000 supported apps by OAuth, API key, user/password, or no sign-in, choose the intended provider, review scopes, and test one harmless action after connecting. 4. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Choose the integration path | Need | Use | | --- | --- | | A supported third-party application | A Composio-managed plugin in Settings → Plugins. | | A custom tool or private service | An MCP server with narrowly scoped capabilities. | | A different AI model provider | Settings → Models, then assign BYOK to the intended agent. | | Messages from an external communication surface | A Channel routed to an agent or bot. | ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/installing-and-connecting-a-plugin title: Installing and connecting a plugin --- # Installing and connecting a plugin Open the catalog, choose the application, start the Composio authorization flow, select the correct account, approve required scopes, and confirm connection status. ![Installing and connecting a plugin in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Open the catalog. 3. Choose the application. 4. Start the Composio authorization flow. 5. Select the correct account. 6. Approve required scopes. 7. Confirm connection status. 8. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/supported-plugin-catalog title: Supported plugin catalog --- # Supported plugin catalog Search by application or capability, inspect available actions and authentication requirements, and choose the narrowest integration that provides the needed operation. ![Supported plugin catalog in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Search by application or capability. 3. Inspect available actions and authentication requirements. 4. Choose the narrowest integration that provides the needed operation. 5. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/using-plugin-tools-in-sessions title: Using plugin tools in tasks --- # Using plugin tools in tasks Attach the plugin tool to an agent, select that agent in a task, request a read-only action first, inspect inputs and results, then approve any intended write action. ![Using plugin tools in tasks in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Attach the plugin tool to an agent. 3. Select that agent in a task, request a read-only action first. 4. Inspect inputs and results. 5. Approve any intended write action. 6. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/managing-connected-accounts title: Managing connected accounts --- # Managing connected accounts Open the plugin’s connections, identify each provider account, test its status, update which agents or automations may use it, and remove stale authorization. ![Managing connected accounts in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Open the plugin’s connections. 3. Identify each provider account. 4. Test its status. 5. Update which agents or automations may use it. 6. Remove stale authorization. 7. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/reconnecting-or-removing-a-plugin title: Reconnecting or removing a plugin --- # Reconnecting or removing a plugin Preserve affected workflow details, reconnect through Composio when authorization expires, retest a harmless action, or disconnect and confirm dependent tools stop working. ![Reconnecting or removing a plugin in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Preserve affected workflow details. 3. Reconnect through Composio when authorization expires, retest a harmless action, or disconnect and confirm dependent tools stop working. 4. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 5. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plugins/plugin-permissions-and-security title: Plugin permissions and security --- # Plugin permissions and security Review provider scopes, Foxora assignment, agent tools, autonomy, and external target together; minimize each layer and audit consequential activity. ![Plugin permissions and security in Foxora v5.0.3](/docs/images/v5.0.3/settings/plugins-catalog.webp) ## Where this lives Open **studio → Plugins** or **Settings → Plugins**. Plugins connect Foxora to third-party applications through Composio. Composio manages the application authorization, and Foxora exposes the approved actions as tools for selected agents, bots, or workflows. > **Safety:** Never paste application passwords or access tokens into a prompt. Review Composio authorization scopes and disconnect unused accounts from Foxora and the provider when necessary. ## Steps 1. Open **studio → Plugins** or **Settings → Plugins**. 2. Review provider scopes, Foxora assignment, agent tools, autonomy, and external target together; minimize each layer and audit consequential activity. 3. **Confirm the result.** Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. ## Choose the right path - Use a plugin for a supported SaaS application. - Use MCP for a custom tool server or internal service. - Use a provider connection or BYOK only for AI model access. ## Confirm it worked - Run a read-only tool first, verify the connected account and returned record, then test any write action in a safe destination with approvals enabled. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/overview title: Overview --- # Overview Connect a messaging surface, define routing and permissions, test inbound and outbound behavior, and operate the channel safely. ![Channels in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Connect a messaging surface. 3. Define routing and permissions. 4. Test inbound and outbound behavior, and operate the channel safely. 5. Open Channels, configure the exact inbound and outbound transport, assign the intended agent or bot, and verify a real received message plus a real reply before treating the channel as ready. 6. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/connecting-a-channel title: Connecting a channel --- # Connecting a channel Choose the provider, authorize the correct account, select the workspace and room, confirm the Foxora identity, and save the connection without broad routing. ![Connecting a channel in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Choose the provider. 3. Authorize the correct account. 4. Select the workspace and room. 5. Confirm the Foxora identity. 6. Save the connection without broad routing. 7. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/routing-messages-to-agents-or-bots title: Routing messages to agents or bots --- # Routing messages to agents or bots Select the source room or event, choose the target bot or agent, define mention or thread behavior, set fallbacks, and test the mapping with a unique message. ![Routing messages to agents or bots in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Select the source room or event. 3. Choose the target bot or agent. 4. Define mention or thread behavior. 5. Set fallbacks. 6. Test the mapping with a unique message. 7. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/channel-permissions title: Channel permissions --- # Channel permissions Restrict allowed rooms and senders, limit the target’s tools, keep high-impact actions approved, and test that unauthorized messages do not trigger work. ![Channel permissions in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Restrict allowed rooms and senders. 3. Limit the target’s tools. 4. Keep high-impact actions approved. 5. Test that unauthorized messages do not trigger work. 6. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/managing-connected-channels title: Managing connected channels --- # Managing connected channels Review connection health, destination mappings, recent traffic, and dependent bots or workflows; pause routing before changing credentials or account ownership. ![Managing connected channels in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Review connection health, destination mappings, recent traffic, and dependent bots or workflows. 3. Pause routing before changing credentials or account ownership. 4. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/channels/channel-troubleshooting title: Channel troubleshooting --- # Channel troubleshooting Check provider authorization, selected workspace and room, event subscriptions, routing rules, bot access, and latest delivery error; then resend one test message. ![Channel troubleshooting in Foxora v5.0.3](/docs/images/v5.0.3/settings/channels.webp) ## Where this lives Open **Settings → Channels**. Channels connect supported messaging surfaces to Foxora bots or agents. Routing rules determine which destination handles an inbound message and where replies are sent. > **Safety:** Channel content is untrusted and may be visible to many people. Restrict senders, rooms, tools, and sensitive outputs; verify the destination before posting or sending. ## Steps 1. Open **Settings → Channels**. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check provider authorization, selected workspace and room, event subscriptions, routing rules, bot access, and latest delivery error. 4. Resend one test message. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. ## Choose the right path - Route repeatable public or team conversations to a bot. - Route controlled internal requests to a specialist agent. - Use a workflow when a message should start a fixed multi-step process. ## Confirm it worked - Send a test message from an allowed location and a blocked location, then verify routing, identity, reply destination, tool activity, and channel audit history. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/overview title: Overview --- # Overview Understand MCP clients, servers, transports, advertised capabilities, configuration, approval boundaries, and the difference from Composio plugins. ![MCP in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Understand MCP clients, servers, transports, advertised capabilities, configuration, approval boundaries, and the difference from Composio plugins. 3. Use **Settings → MCP Server** to enable or inspect Foxora’s MCP endpoint and exposed capabilities. For external MCP servers, configure the trusted transport and credentials through the supported tool or plugin flow. 4. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/adding-an-mcp-server title: Adding an MCP server --- # Adding an MCP server Create a server entry, choose local or remote transport, enter the command or endpoint, add required dedicated credentials, connect, and inspect discovered capabilities. ![Adding an MCP server in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Create a server entry. 3. Choose local or remote transport. 4. Enter the command or endpoint. 5. Add required dedicated credentials. 6. Connect. 7. Inspect discovered capabilities. 8. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/configuring-mcp-servers title: Configuring MCP servers --- # Configuring MCP servers Set transport, arguments, environment, authentication, timeout, and allowed capabilities explicitly; save, reconnect, and confirm the effective configuration. ![Configuring MCP servers in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Set transport, arguments, environment, authentication, timeout, and allowed capabilities explicitly. 3. Save. 4. Reconnect. 5. Confirm the effective configuration. 6. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/using-mcp-tools-in-sessions title: Using MCP tools in tasks --- # Using MCP tools in tasks Enable selected MCP tools for an agent, open a test task, request a read-only operation, inspect the exact server and arguments, and verify the returned data. ![Using MCP tools in tasks in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Enable selected MCP tools for an agent. 3. Open a test task, request a read-only operation. 4. Inspect the exact server and arguments. 5. Verify the returned data. 6. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/using-foxora-as-an-mcp-server title: Using Foxora as an MCP server --- # Using Foxora as an MCP server Enable the Foxora MCP server surface, choose exposed capabilities and authentication, add it to the client, then test discovery and one bounded call. ![Using Foxora as an MCP server in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Enable the Foxora MCP server surface. 3. Choose exposed capabilities and authentication. 4. Add it to the client. 5. Test discovery and one bounded call. 6. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Client configuration shape Use the URL shown by **Settings → MCP Server**. A client configuration follows this shape: ```json { "mcpServers": { "foxora": { "url": "/mcp" } } } ``` Foxora shows the exact exposed tool list beside the configuration. Review that list in the client before approving calls. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/managing-and-removing-servers title: Managing and removing servers --- # Managing and removing servers Review dependent agents and workflows, disable the server, confirm tools disappear from new tasks, then remove its configuration and revoke credentials. ![Managing and removing servers in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Review dependent agents and workflows. 3. Disable the server. 4. Confirm tools disappear from new tasks. 5. Remove its configuration and revoke credentials. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/mcp/mcp-security-and-troubleshooting title: MCP security and troubleshooting --- # MCP security and troubleshooting Check process or endpoint reachability, transport logs, authentication, schema compatibility, and capability discovery; keep privileges minimal while isolating the failing layer. ![MCP security and troubleshooting in Foxora v5.0.3](/docs/images/v5.0.3/settings/mcp-server.webp) ## Where this lives Open **Settings → MCP Server**. Model Context Protocol connects Foxora to external tool and resource servers. Foxora can consume configured MCP servers and can expose supported Foxora capabilities to an MCP client. > **Safety:** An MCP server can run code or access external systems with its own privileges. Trust the publisher, pin configuration where possible, and never place secrets in arguments or prompts. ## Steps 1. Open **Settings → MCP Server**. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check process or endpoint reachability, transport logs, authentication, schema compatibility, and capability discovery. 4. Keep privileges minimal while isolating the failing layer. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. ## Choose the right path - Use local stdio for a server installed on this device. - Use remote transport for an authenticated managed endpoint. - Use a Composio plugin when the target SaaS application is already supported. ## Confirm it worked - Call one read-only tool with controlled input, inspect its result and server logs, then verify disconnecting the server removes those capabilities from new work. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/overview title: Overview --- # Overview Choose between Foxora-managed models and agent-level BYOK, then select a model or processing tier that fits the task and budget. ![Models in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Choose between Foxora-managed models and agent-level BYOK. 3. Select a model or processing tier that fits the task and budget. 4. Use **Settings → Models** to inspect managed routing and supported provider configuration. Select the intended agent or crew tier, keep provider keys in protected fields, and verify the effective model through a representative task. 5. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Managed models and BYOK | Configuration | Where it is selected | Usage source | | --- | --- | --- | | Foxora-managed | Choose the managed model or processing tier for the agent | Uses the subscription’s Foxora credits. | | BYOK | Add a supported provider key to the particular agent and select its provider model | Uses that provider account and its limits or charges. | Foxora-managed access remains the default. A BYOK choice on one agent does not need to change every other agent. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/connecting-an-ai-provider title: Connecting an AI provider --- # Connecting an AI provider Open provider settings, choose a supported provider, enter the key in the protected field, validate the connection, and assign it only to the intended agent. ![Connecting an AI provider in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Open provider settings. 3. Choose a supported provider. 4. Enter the key in the protected field. 5. Validate the connection. 6. Assign it only to the intended agent. 7. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/bring-your-own-key title: Bring Your Own Key --- # Bring Your Own Key Create or edit an agent, switch its model source to BYOK, select the provider and model, add or choose the key, save, and verify provider-side usage. ![Bring Your Own Key in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Create or edit an agent. 3. Switch its model source to BYOK. 4. Select the provider and model. 5. Add or choose the key. 6. Save. 7. Verify provider-side usage. 8. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/selecting-models title: Selecting models --- # Selecting models Filter compatible models, compare capability, latency, context, and provider pricing, select one for the agent, and test with a representative request. ![Selecting models in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Filter compatible models. 3. Compare capability, latency, context, and provider pricing. 4. Select one for the agent. 5. Test with a representative request. 6. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/processing-tiers title: Processing tiers --- # Processing tiers Use Fast for routine work, Premium for stronger general reasoning, or Max for the hardest tasks; confirm availability under the active subscription before running. ![Processing tiers in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Use Fast for routine work, Premium for stronger general reasoning, or Max for the hardest tasks. 3. Confirm availability under the active subscription before running. 4. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/default-and-per-session-models title: Default and per-task models --- # Default and per-task models Set a stable default through the agent configuration, choose the appropriate agent or crew in each task, and confirm the effective model before sending expensive work. ![Default and per-task models in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Set a stable default through the agent configuration. 3. Choose the appropriate agent or crew in each task. 4. Confirm the effective model before sending expensive work. 5. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/models-providers/availability-cost-and-limits title: Availability, cost, and limits --- # Availability, cost, and limits Check plan credits for managed access or provider limits for BYOK, inspect current usage and rate-limit messages, and reduce scope or change models when needed. ![Availability, cost, and limits in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open **Settings → Models**. Foxora provides managed model access through subscription credits and also supports BYOK. A BYOK provider key is assigned to a particular agent, so that agent uses the selected provider and model instead of Foxora-managed credits for those calls. > **Safety:** Add API keys only through the dedicated agent or provider controls. Never include keys in agent instructions, task prompts, project files, screenshots, or support bundles. ## Steps 1. Open **Settings → Models**. 2. Check plan credits for managed access or provider limits for BYOK. 3. Inspect current usage and rate-limit messages. 4. Reduce scope or change models when needed. 5. **Confirm the result.** Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. ## Choose the right path - Use managed models for the simplest setup and Foxora credit accounting. - Use BYOK for a supported provider model billed by that provider. - Choose Fast, Premium, or Max processing according to task complexity and plan access. ## Confirm it worked - Run a known prompt with the configured agent, inspect the effective model and usage record, and confirm whether Foxora credits or the provider account was charged. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/overview title: Overview --- # Overview Navigate each settings category, distinguish global defaults from task or agent configuration, and verify changes in the affected surface. ![Settings in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/settings-appearance.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Navigate each settings category. 3. Distinguish global defaults from task or agent configuration. 4. Verify changes in the affected surface. 5. Change one section at a time, distinguish account-wide defaults from task or agent configuration, save where required, and retest the affected surface. 6. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/appearance-settings title: Appearance settings --- # Appearance settings Choose theme, accent, density, text scale, and motion preferences; apply them and check conversation, editor, terminal, and Stage readability. ![Appearance settings in Foxora Desktop v5.0.3](/docs/images/v5.0.3/desktop/settings-appearance.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Choose theme, accent, density, text scale, and motion preferences. 3. Apply them and check conversation, editor, terminal, and Stage readability. 4. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/voice-settings title: Voice settings --- # Voice settings Select microphone and speaker devices, grant OS access, set voice behavior, run an input and output test, and disable listening when unused. ![Voice settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/voice.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Select microphone and speaker devices. 3. Grant OS access. 4. Set voice behavior. 5. Run an input and output test. 6. Disable listening when unused. 7. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/browser-settings title: Browser settings --- # Browser settings Choose managed or existing-browser mode, authorize the required bridge or permissions, test a harmless page, and confirm the active browser context. ![Browser settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/browser.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Choose managed or existing-browser mode. 3. Authorize the required bridge or permissions. 4. Test a harmless page. 5. Confirm the active browser context. 6. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/plugin-and-channel-settings title: Plugin and channel settings --- # Plugin and channel settings Open the relevant connection, review provider account and scopes, test status, update assignments, and disconnect only after checking dependent bots or workflows. ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Open the relevant connection. 3. Review provider account and scopes. 4. Test status. 5. Update assignments. 6. Disconnect only after checking dependent bots or workflows. 7. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/brand-settings title: Brand settings --- # Brand settings Set approved name, logo, colors, and presentation defaults, preview them on the intended bot or shared surface, and verify accessibility and ownership rights. ![Brand settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/brand.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Set approved name, logo, colors, and presentation defaults, preview them on the intended bot or shared surface. 3. Verify accessibility and ownership rights. 4. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/permission-settings title: Permission settings --- # Permission settings Review filesystem, terminal, browser, computer-use, microphone, and notification access; grant the minimum needed and retest after revocation. ![Permission settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Review filesystem, terminal, browser, computer-use, microphone, and notification access. 3. Grant the minimum needed and retest after revocation. 4. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/data-and-system-settings title: Data and system settings --- # Data and system settings Inspect local storage and runtime paths, export required data, adjust supported system options, and use reset or clear actions only after reviewing their exact scope. ![Data and system settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/data.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Inspect local storage and runtime paths. 3. Export required data. 4. Adjust supported system options. 5. Use reset or clear actions only after reviewing their exact scope. 6. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/keyboard-settings title: Keyboard settings --- # Keyboard settings Open shortcut mappings, identify conflicts, change the binding, save, and test it in the relevant workspace context. ![Keyboard settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/keyboard.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Open shortcut mappings. 3. Identify conflicts. 4. Change the binding. 5. Save. 6. Test it in the relevant workspace context. 7. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/settings/about-and-update-settings title: About and update settings --- # About and update settings Check Desktop and runtime versions, read release information, save active work, install the signed update, restart, and verify a small task. ![About and update settings in Foxora v5.0.3](/docs/images/v5.0.3/settings/about.webp) ## Where this lives Open **Settings** and select the named section in the left rail. Settings contains Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About. > **Safety:** Permission, data, connection, and update settings can affect security or availability. Preserve diagnostics and exports before reset or removal actions. ## Steps 1. Open **Settings** and select the named section in the left rail. 2. Check Desktop and runtime versions. 3. Read release information. 4. Save active work. 5. Install the signed update. 6. Restart. 7. Verify a small task. 8. **Confirm the result.** Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. ## Choose the right path - Use workspace controls for temporary layout changes. - Use Settings for durable application defaults. - Use an agent or task control when behavior should apply only to that configuration. ## Confirm it worked - Reopen the relevant screen or restart only when prompted, then confirm the effective value and verify that existing tasks, agents, and automations still behave as intended. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/overview title: Overview --- # Overview Create and protect an account, sign clients in through the browser, manage profile and devices, recover access, and understand local-data boundaries. ![Account & Authentication in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Create and protect an account, sign clients in through the browser, manage profile and devices, recover access. 3. Understand local-data boundaries. 4. Use a unique password or supported identity provider, verify the browser sign-in handoff, review authorized devices, and keep recovery access current. 5. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/creating-an-account title: Creating an account --- # Creating an account Open the official sign-up page, enter the requested account details, verify the email or identity-provider flow, choose a plan when prompted, and complete the profile. ![Creating an account in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Open the official sign-up page. 3. Enter the requested account details. 4. Verify the email or identity-provider flow. 5. Choose a plan when prompted. 6. Complete the profile. 7. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/signing-in title: Signing in --- # Signing in Start sign-in from Foxora, inspect the browser domain and requested account, authenticate, approve the handoff, and return to confirm the signed-in identity. ![Signing in in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Start sign-in from Foxora. 3. Inspect the browser domain and requested account, authenticate. 4. Approve the handoff, and return to confirm the signed-in identity. 5. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/resetting-your-password title: Resetting your password --- # Resetting your password Open the official reset flow, enter the account email, use the newest reset message, choose a unique password, then sign in and review authorized devices. ![Resetting your password in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Open the official reset flow. 3. Enter the account email. 4. Use the newest reset message. 5. Choose a unique password. 6. Sign in and review authorized devices. 7. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/profile-settings title: Profile settings --- # Profile settings Open Account settings, edit supported identity or profile fields, save once, and confirm the updated information in another Foxora surface. ![Profile settings in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Open Account settings. 3. Edit supported identity or profile fields. 4. Save once. 5. Confirm the updated information in another Foxora surface. 6. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/authorized-devices title: Authorized devices --- # Authorized devices Review device name, platform, location or activity metadata, identify the current installation, and revoke any device you no longer recognize or control. ![Authorized devices in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Review device name, platform, location or activity metadata. 3. Identify the current installation. 4. Revoke any device you no longer recognize or control. 5. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/signing-out-of-other-devices title: Signing out of other devices --- # Signing out of other devices Select the exact remote device or use the all-other-devices action, confirm revocation, then change credentials if access may have been compromised. ![Signing out of other devices in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Select the exact remote device or use the all-other-devices action. 3. Confirm revocation. 4. Change credentials if access may have been compromised. 5. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 6. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/account-authentication/deleting-your-account title: Deleting your account --- # Deleting your account Export required account and local data, cancel dependent subscriptions or connections, review the permanent deletion scope, confirm once, and separately clear local data on each device if desired. ![Deleting your account in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the account menu or the Foxora sign-in screen. One Foxora account provides authentication, subscription, credits, connected web services, and access across supported clients. Desktop and CLI share a device’s local runtime, but its sessions and Memory Den records do not sync to the web app or another device merely because the same account signs in. > **Safety:** Never approve a sign-in you did not start. Signing out or deleting an account is separate from deleting local sessions and Memory Den data on each device. ## Steps 1. Open the account menu or the Foxora sign-in screen. 2. Export required account and local data. 3. Cancel dependent subscriptions or connections. 4. Review the permanent deletion scope. 5. Confirm once, and separately clear local data on each device if desired. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. ## Choose the right path - Create one account for a new identity. - Sign in to connect an installed client to an existing account. - Revoke another device when it is lost, shared, or no longer trusted. ## Confirm it worked - Open the account profile after sign-in, confirm the expected email and plan, and review the authorized-device list for the current client. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/overview title: Overview --- # Overview Choose among Free, Starter, Pro, and Max, understand monthly Foxora credits, distinguish BYOK billing, and monitor usage before limits interrupt work. ![Plans, Credits & Usage in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-billing.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Choose among Free, Starter, Pro, and Max. 3. Understand monthly Foxora credits. 4. Distinguish BYOK billing. 5. Monitor usage before limits interrupt work. 6. Confirm the active plan, current credit window, expected workload, managed or BYOK model source, billing cycle, and renewal date before changing a subscription. 7. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/plans-and-pricing title: Plans and pricing --- # Plans and pricing Compare Free, Starter, Pro, and Max by included credits and required capabilities, select monthly or available longer-term billing, and confirm the checkout summary before purchase. ![Plans and pricing in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-billing.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Compare Free, Starter, Pro, and Max by included credits and required capabilities. 3. Select monthly or available longer-term billing. 4. Confirm the checkout summary before purchase. 5. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Included monthly credits | Plan | Foxora credits | | --- | ---: | | Free | 200 | | Starter | 2,000 | | Pro | 5,000 | | Max | 20,000 | Use the live Billing screen for the current price, billing cycle, taxes, and checkout total. BYOK provider usage is billed and limited separately by that provider. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/understanding-credits title: Understanding credits --- # Understanding credits Identify whether an agent uses managed access or BYOK, inspect the managed run’s credit usage, compare task complexity and tier, and use the usage window to plan remaining work. ![Understanding credits in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-usage.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Identify whether an agent uses managed access or BYOK. 3. Inspect the managed run’s credit usage. 4. Compare task complexity and tier. 5. Use the usage window to plan remaining work. 6. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/viewing-usage title: Viewing usage --- # Viewing usage Open Usage or run `foxora usage`, review weekly and monthly windows, filter or inspect recent activity where available, and investigate unexpected changes before more runs. ![Viewing usage in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-usage.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Open Usage or run `foxora usage`. 3. Review weekly and monthly windows. 4. Filter or inspect recent activity where available, and investigate unexpected changes before more runs. 5. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/usage-and-rate-limits title: Usage and rate limits --- # Usage and rate limits Read whether the notice concerns credits, a time window, concurrency, or a BYOK provider limit; wait, reduce load, change configuration, or upgrade only after identifying the limit. ![Usage and rate limits in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-usage.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Read whether the notice concerns credits, a time window, concurrency, or a BYOK provider limit. 3. Wait. 4. Reduce load. 5. Change configuration, or upgrade only after identifying the limit. 6. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/managing-a-subscription title: Managing a subscription --- # Managing a subscription Open Billing, confirm the current plan and renewal date, select the intended plan or cycle, review the effective date and total, and save the confirmation. ![Managing a subscription in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-billing.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Open Billing. 3. Confirm the current plan and renewal date. 4. Select the intended plan or cycle. 5. Review the effective date and total. 6. Save the confirmation. 7. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/updating-billing-information title: Updating billing information --- # Updating billing information Open the secure billing portal, select the subscription, update the supported payment and billing fields, save, and confirm the next invoice uses the intended details. ![Updating billing information in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-billing.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Open the secure billing portal. 3. Select the subscription. 4. Update the supported payment and billing fields. 5. Save. 6. Confirm the next invoice uses the intended details. 7. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/plans-credits-usage/cancelling-a-subscription title: Cancelling a subscription --- # Cancelling a subscription Review the access end date, unused credits, exports, active automations, and BYOK keys; confirm cancellation once and verify the account shows the expected future plan state. ![Cancelling a subscription in Foxora v5.0.3](/docs/images/v5.0.3/settings/account-billing.webp) ## Where this lives Open **Settings → Account** and select Overview, Billing, or Usage. Foxora subscriptions include monthly managed-model credits: Free includes 200, Starter 2,000, Pro 5,000, and Max 20,000. BYOK remains optional and provider-billed usage is managed separately from Foxora credits. > **Safety:** Credits are usage units, not money stored in the account. BYOK calls can incur separate provider charges; subscription cancellation does not revoke provider keys automatically. ## Steps 1. Open **Settings → Account** and select Overview, Billing, or Usage. 2. Review the access end date, unused credits, exports, active automations, and BYOK keys. 3. Confirm cancellation once and verify the account shows the expected future plan state. 4. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 5. **Confirm the result.** Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. ## Choose the right path - Free — 200 Foxora credits. - Starter — 2,000 Foxora credits. - Pro — 5,000 Foxora credits. - Max — 20,000 Foxora credits. ## Confirm it worked - Refresh Usage and Billing after the change, confirm the plan and credit allowance, and compare a small test run with its usage record. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/overview title: Overview --- # Overview Apply layered security to accounts, projects, autonomy, tools, connections, local storage, secrets, exports, and deletion. ![Security & Privacy in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Apply layered security to accounts, projects, autonomy, tools, connections, local storage, secrets, exports, and deletion. 3. Inventory the data and external systems involved, use the narrowest project and permissions, keep consequential actions approved, and store secrets only in dedicated controls. 4. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/permission-and-autonomy-security title: Permission and autonomy security --- # Permission and autonomy security Classify task risk, choose Ask Permissions, Accept Edits, Auto, or Bypass Permissions, review tool targets, and lower autonomy after the bounded task. ![Permission and autonomy security in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Classify task risk. 3. Choose Ask Permissions, Accept Edits, Auto, or Bypass Permissions. 4. Review tool targets, and lower autonomy after the bounded task. 5. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/local-data-storage title: Local data storage --- # Local data storage Review the device locations and application controls for sessions, Memory Den, attachments, logs, and settings; secure the OS account, backups, and disk. ![Local data storage in Foxora v5.0.3](/docs/images/v5.0.3/settings/data.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Review the device locations and application controls for sessions, Memory Den, attachments, logs, and settings; secure the OS account, backups, and disk. 3. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/memory-privacy title: Memory privacy --- # Memory privacy Inspect saved local memories and provenance, remove sensitive or outdated entries, test recall, and remember that clearing a session does not necessarily clear its durable memories. ![Memory privacy in Foxora v5.0.3](/docs/images/v5.0.3/settings/data.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Inspect saved local memories and provenance. 3. Remove sensitive or outdated entries. 4. Test recall, and remember that clearing a session does not necessarily clear its durable memories. 5. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/authentication-and-session-security title: Authentication and task security --- # Authentication and task security Use the official browser flow, verify account identity, review authorized devices, revoke unknown sessions, and reset credentials after suspicious activity. ![Authentication and task security in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Use the official browser flow. 3. Verify account identity. 4. Review authorized devices. 5. Revoke unknown sessions, and reset credentials after suspicious activity. 6. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/managing-secrets-and-api-keys title: Managing secrets and API keys --- # Managing secrets and API keys Add provider keys only to dedicated agent or connection fields, restrict the associated agent and tools, rotate exposed keys at the provider, and retest after replacement. ![Managing secrets and API keys in Foxora v5.0.3](/docs/images/v5.0.3/settings/models.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Add provider keys only to dedicated agent or connection fields. 3. Restrict the associated agent and tools. 4. Rotate exposed keys at the provider, and retest after replacement. 5. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/security-privacy/exporting-or-deleting-data title: Exporting or deleting data --- # Exporting or deleting data Choose account data, local sessions, local Memory Den, or application settings separately, export what must be retained, verify the archive, then confirm only the intended deletion. ![Exporting or deleting data in Foxora v5.0.3](/docs/images/v5.0.3/settings/data.webp) ## Where this lives Open the relevant section under **Settings → Permissions** or **Settings → Data**. Foxora security is layered across account authentication, local project access, session autonomy, agent tools, browser and computer control, connected services, and local data storage. > **Safety:** No autonomy mode can make an external action reversible. Use separate test environments, least privilege, independent verification, and prompt incident response for suspected credential exposure. ## Steps 1. Open the relevant section under **Settings → Permissions** or **Settings → Data**. 2. Choose account data, local sessions, local Memory Den, or application settings separately. 3. Export what must be retained. 4. Verify the archive. 5. Confirm only the intended deletion. 6. Review what will be retained, revoked, detached, stopped, or permanently deleted; export anything you may need later. 7. **Confirm the result.** Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. ## Choose the right path - Use local sessions and Memory Den for device-resident work. - Use Composio authorization for supported plugin accounts. - Use an agent’s dedicated BYOK field for model-provider keys. ## Confirm it worked - Review the transcript and external audit records, inspect authorized devices and connections, and test that revoked access no longer works. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/cli/overview title: Overview --- # Overview The Foxora CLI is the agent terminal — a fast, native command-line client that runs the same crews of agents as the desktop app, straight from your shell. Type `foxora` to open the interactive UI, or drive it headlessly with commands such as `foxora run`, `foxora sessions`, and `foxora doctor`. ## What the CLI is foxora is the command-line client for Foxora. On first run it provisions the Foxora Runtime (the agent engine, Memory Den, and a workspace) into a per-user cache and runs session data locally. Foxora-managed model calls route through the Foxora gateway; BYOK remains an optional agent-level model configuration. Run with no arguments and it opens a full-screen terminal UI. Pass a subcommand and it runs headless, printing to stdout like any other CLI — perfect for scripts, CI, and quick one-shots. ## When to use it - You live in the terminal — drive agents without leaving your shell. - One-shot prompts — `foxora run "summarize this repo"` streams a reply and exits. - Scripting & CI — the headless subcommands print plain text you can pipe. - Headless boxes — no desktop app required; the CLI provisions its own runtime. The CLI uses the same Foxora account as the other clients. On one computer, Desktop and CLI share the device runtime and `~/.foxora` data home, including local sessions and Memory Den. The web app uses its own cloud workspace; signing in to the same account does not copy the device’s local sessions, files, or memories into the browser. > **This track targets v5.0.3** Commands, flags, and output shown here describe the foxora CLI as of v5.0.3. Check your version any time with `foxora --version`, and keep it current with `foxora update`. ## Core concepts ### Runtime The engine + memory + workspace bundle the CLI installs into `~/.foxora/runtime`. `foxora setup` provisions it; `foxora doctor` tells you whether it’s healthy. ### Crew (tier) A crew is a tier of agents — fast, pro, or max — that plan, act, and verify their own work. You can pick a tier per run; otherwise the CLI routes for you. ### Session Each conversation is a session stored by the local Foxora runtime. Desktop and CLI connect to that same runtime on the device, so both can access its local sessions and Memory Den. This data does not automatically sync to the web app or another device. List the available local sessions with `foxora sessions`. --- source: https://docs.foxora.ai/cli/installing-the-cli title: Installing the CLI --- # Installing the CLI Install the Foxora CLI on macOS, Linux, or Windows, then provision its runtime on first use. ## macOS and Linux ```bash curl -fsSL https://foxora.ai/install-cli.sh | sh ``` Open a new terminal after installation so your updated `PATH` is available. ## Windows Run the installer from PowerShell: ```powershell irm https://foxora.ai/install.ps1 | iex ``` Open a new PowerShell window after the installer finishes. ## Verify and provision ```bash foxora --version foxora setup foxora login foxora doctor ``` The documentation targets CLI v5.0.3. `foxora --version` reports the version actually installed on your machine, and `foxora update` installs a newer release when one is available. > **Install safely:** Use only the scripts served from `https://foxora.ai`. If you want to inspect a script first, download or print it before piping it to your shell. --- source: https://docs.foxora.ai/cli/authentication-and-setup title: Authentication and setup --- # Authentication and setup Provision the runtime once, then authenticate the CLI with your Foxora account. ## Set up the runtime ```bash foxora setup ``` This prepares the engine, Memory Den, and Bun runtime used by the CLI. ## Sign in ```bash foxora login ``` Foxora opens the browser for authentication. After the browser flow completes, return to the terminal and verify the active account: ```bash foxora whoami ``` ## Sign out ```bash foxora logout ``` If setup or sign-in fails, run `foxora doctor` and use its engine, runtime, and authentication results to identify the failing layer. --- source: https://docs.foxora.ai/cli/interactive-cli-usage title: Interactive CLI usage --- # Interactive CLI usage Run `foxora` without a subcommand to open the full-screen interactive terminal UI. ## Start an interactive session ```bash foxora ``` The interactive UI is intended for supervised conversations and ongoing sessions. Use `foxora run` instead when a script needs one prompt, one printed result, and an exit code. ## Before opening the UI ```bash foxora setup foxora login foxora doctor ``` - `setup` provisions the local runtime. - `login` opens the secure browser sign-in flow. - `doctor` checks the engine, runtime, and authentication if the UI cannot start. Use `foxora --remote` when you explicitly want to attach to an already-running engine without allowing the CLI to spawn one. --- source: https://docs.foxora.ai/cli/headless-runs title: Headless runs --- # Headless runs Use `foxora run` for a one-shot request that prints its reply to stdout and exits. ## Run with the default crew ```bash foxora run "summarize this repository" ``` ## Select a crew tier Pass `fast`, `pro`, or `max` immediately after `run`: ```bash foxora run fast -n "fix the failing test" foxora run pro -n "review this architecture" foxora run max -n "analyze the hardest unresolved issue" ``` The prompt can be supplied as the remaining quoted argument or explicitly with `-n` / `--prompt`. ## Shell usage Headless output is plain text, so it can be redirected or captured with normal shell tools: ```bash foxora run fast -n "review this diff" > review.txt ``` Use the process exit code to detect failure. Foxora CLI v5.0.3 does not expose a `--json` option. --- source: https://docs.foxora.ai/cli/command-reference title: Command reference --- # Command reference Foxora CLI v5.0.3 supports the commands below. Run `foxora --help` on your installed version as the final authority. ## Commands | Command | Purpose | | --- | --- | | `foxora` | Open the interactive terminal UI. | | `foxora setup` | Set up the runtime: engine, Memory Den, and Bun. | | `foxora update` | Update Foxora and its runtime to the latest release. | | `foxora login` | Sign in through the browser. | | `foxora logout` | Sign out. | | `foxora run [crew] -n ` | Run one prompt, print the reply, and exit. | | `foxora usage` | Show weekly and monthly credit usage. | | `foxora doctor` | Diagnose the engine, runtime, and authentication. | | `foxora models` | List available tiers and models. | | `foxora whoami` | Show the signed-in account. | | `foxora sessions` | List sessions. | | `foxora crews` | List crews. | | `foxora agents` | List agents. | | `foxora --help` | Show command usage, options, and environment variables. | ## Options | Option | Purpose | | --- | --- | | `-n, --prompt ` | Supply the prompt for `foxora run`. | | `--remote` | Attach to a running engine without spawning one. | | `-h, --help` | Show command help. | | `-v, --version` | Print the installed CLI version. | ## Run examples ```bash foxora run "summarize this repo" foxora run fast -n "fix the failing test" ``` The optional crew argument accepts `fast`, `pro`, or `max`. If you omit it, Foxora chooses the default routing behavior. > **Exact names matter:** The list commands are plural: `foxora sessions`, `foxora crews`, and `foxora agents`. Commands such as `foxora crew`, `foxora queue`, `foxora recall`, and `foxora org` are not part of v5.0.3. --- source: https://docs.foxora.ai/cli/configuration-and-updates title: Configuration and updates --- # Configuration and updates Use the installed CLI to inspect its version, update the runtime, and override service endpoints when required. ## Check and update ```bash foxora --version foxora update foxora doctor ``` `foxora update` updates both the CLI and its managed runtime. Run `foxora doctor` afterward if the engine, authentication, or runtime does not come back healthy. ## Environment variables | Variable | Default | Purpose | | --- | --- | --- | | `FOXORA_ENGINE_URL` | `https://engine.foxora.ai` | Engine endpoint. | | `FOXORA_GATEWAY_URL` | `https://api.foxora.ai` | Gateway endpoint. | | `FOXORA_RESOURCE_ID` | Current account ID | Memory bucket identifier. | Use `--remote` when attaching to an existing engine: ```bash foxora --remote ``` Only override service URLs in a trusted development or managed deployment environment. --- source: https://docs.foxora.ai/cli/cli-troubleshooting title: CLI troubleshooting --- # CLI troubleshooting Use the CLI’s built-in checks before resetting configuration or reinstalling the runtime. ## `foxora` is not found Open a new terminal after installation, then run: ```bash foxora --version ``` If the command is still unavailable, confirm that the installer’s binary directory is included in your user `PATH`. ## The runtime does not start ```bash foxora setup foxora doctor ``` `setup` provisions the engine, Memory Den, and Bun runtime. `doctor` reports which of the engine, runtime, or authentication checks failed. ## Authentication fails ```bash foxora login foxora whoami ``` Complete the browser flow, then use `whoami` to confirm which account is active. Use `foxora logout` before signing into a different account. ## Unknown command ```bash foxora --help ``` CLI v5.0.3 uses plural list commands: `foxora sessions`, `foxora crews`, and `foxora agents`. One-shot prompts must use `foxora run`; a bare quoted prompt is not a command. ## Version or update problems ```bash foxora --version foxora update foxora doctor ``` Record the exact command, version, and diagnostic output before contacting support. Avoid deleting `~/.foxora` as a troubleshooting step because it may contain sessions, runtime state, and Memory Den data. --- source: https://docs.foxora.ai/api/overview title: Overview --- # Overview Make a first authenticated API request, verify the response and usage, and use the endpoint-specific pages for production handling. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Make a first authenticated API request. 3. Verify the response and usage. 4. Use the endpoint-specific pages for production handling. 5. Obtain an account access token through the supported sign-in flow, keep it out of source control and logs, select a returned model identifier, send JSON with explicit timeouts, and handle every documented status code. 6. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/authentication title: Authentication --- # Authentication Use a signed-in Foxora account access token as a Bearer token over HTTPS, test `/v1/me`, and sign out or revoke the session after suspected exposure. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Use a signed-in Foxora account access token as a Bearer token over HTTPS. 3. Test `/v1/me`, and sign out or revoke the session after suspected exposure. 4. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Authentication header Store the signed-in account access token outside source code and send it as a Bearer token: ```bash curl https://api.foxora.ai/v1/me \ -H "Authorization: Bearer $FOXORA_ACCESS_TOKEN" ``` > **Developer API status:** Self-service long-lived user API keys are not exposed in Foxora v5.0.3. Do not copy internal service credentials from application storage. Use a short-lived account access token only in a trusted server-side or local development environment. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/chat-completions-api title: Chat Completions API --- # Chat Completions API Fetch a valid model identifier, construct the messages payload, send `POST /v1/chat/completions`, parse the response and usage, and handle streaming only as documented. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Fetch a valid model identifier. 3. Construct the messages payload. 4. Send `POST /v1/chat/completions`, parse the response and usage, and handle streaming only as documented. 5. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Minimal request Get a currently available model identifier from `/v1/models`, then send an OpenAI-compatible request: ```bash curl https://api.foxora.ai/v1/chat/completions \ -H "Authorization: Bearer $FOXORA_ACCESS_TOKEN" \ -H "Content-Type: application/json" \ -d '{ "model": "", "messages": [ { "role": "user", "content": "Summarize this in three bullets." } ] }' ``` Read the returned choice and usage data, and preserve the request identifier when reporting an API failure. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/account-endpoint title: Account endpoint --- # Account endpoint Send authenticated `GET /v1/me`, validate the status and account fields, and use the result to confirm the token belongs to the expected account. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Send authenticated `GET /v1/me`. 3. Validate the status and account fields. 4. Use the result to confirm the token belongs to the expected account. 5. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Request ```bash curl https://api.foxora.ai/v1/me \ -H "Authorization: Bearer $FOXORA_ACCESS_TOKEN" ``` Use this endpoint as an authentication and account check before enabling plan-dependent features. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/usage-endpoint title: Usage endpoint --- # Usage endpoint Send authenticated `GET /v1/me/usage`, inspect weekly and monthly windows and plan context, and compare the response before and after a controlled request. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Send authenticated `GET /v1/me/usage`. 3. Inspect weekly and monthly windows and plan context. 4. Compare the response before and after a controlled request. 5. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Request ```bash curl "https://api.foxora.ai/v1/me/usage?days=30" \ -H "Authorization: Bearer $FOXORA_ACCESS_TOKEN" ``` Choose a supported day range, validate the returned interval, and keep Foxora-managed usage separate from BYOK provider billing. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/models-endpoint title: Models endpoint --- # Models endpoint Send `GET /v1/models`, select a currently returned identifier compatible with the desired request, and refresh discovery when availability changes. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Send `GET /v1/models`. 3. Select a currently returned identifier compatible with the desired request. 4. Refresh discovery when availability changes. 5. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Request ```bash curl https://api.foxora.ai/v1/models ``` Use an identifier returned for the current account. Model availability can change, so production clients should refresh discovery instead of relying on a hardcoded catalog. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/errors-and-status-codes title: Errors and status codes --- # Errors and status codes Log the status, request identifier, safe response details, and retryability; correct invalid requests, reauthenticate on authorization errors, and back off on transient limits. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Log the status, request identifier, safe response details, and retryability. 3. Correct invalid requests, reauthenticate on authorization errors, and back off on transient limits. 4. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Error handling guide | Status | Meaning | Client action | | --- | --- | --- | | 400 | Invalid request | Fix the payload; do not retry it unchanged. | | 401 | Missing, expired, or invalid authentication | Refresh the account session and replace the access token. | | 403 | Account or feature not permitted | Verify account, plan, and key scope. | | 404 | Endpoint or resource not found | Check the API path and resource identifier. | | 429 | Rate or usage limit reached | Respect retry guidance and reduce request rate. | | 5xx | Temporary service failure | Retry only idempotent work with bounded backoff and jitter. | ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/api/rate-limits-and-versioning title: Rate limits and versioning --- # Rate limits and versioning Read rate-limit headers and error bodies, use bounded exponential backoff with jitter for retryable responses, cap concurrency, and pin the documented API version behavior. ## Where this lives Use a trusted local or server-side API client. The Foxora API provides OpenAI-compatible chat completions plus account, usage, and public model-discovery endpoints. Authenticated routes currently accept a signed-in Foxora account access token. > **Safety:** Treat the complete account access token as a secret. Do not paste it into prompts, commit it to a repository, expose it in client-side bundles, or include it in logs, screenshots, and support messages. ## Steps 1. Use a trusted local or server-side API client. 2. Read rate-limit headers and error bodies. 3. Use bounded exponential backoff with jitter for retryable responses, cap concurrency. 4. Pin the documented API version behavior. 5. **Confirm the result.** Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. ## Choose the right path - Use `/v1/chat/completions` for OpenAI-compatible generation. - Use `/v1/me` and `/v1/me/usage` for account and usage state. - Use `/v1/models` to discover model identifiers instead of hardcoding assumptions. ## Confirm it worked - Inspect HTTP status, response headers, request identifiers, usage fields, and the parsed response; cross-check account or usage endpoints after a small request. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/overview title: Overview --- # Overview Diagnose Foxora systematically by isolating the failing layer, preserving evidence, applying one reversible fix, and verifying with the same test. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Diagnose Foxora systematically by isolating the failing layer, preserving evidence, applying one reversible fix, and verifying with the same test. 3. Reproduce once with the smallest safe input, capture exact time and error, preserve logs and project state, check versions and service status, then change one variable. 4. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/installation-and-startup-problems title: Installation and startup problems --- # Installation and startup problems Confirm OS and architecture support, installer source and version, disk and security prompts, runtime status, then repair or reinstall without clearing local data first. ![Installation and startup problems in Foxora v5.0.3](/docs/images/v5.0.3/settings/system.webp) ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Confirm OS and architecture support, installer source and version, disk and security prompts, runtime status. 4. Repair or reinstall without clearing local data first. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/account-and-sign-in-problems title: Account and sign-in problems --- # Account and sign-in problems Check connectivity, browser domain, account identity, clock, authorized devices, and subscription state; repeat the browser handoff and capture the exact failure. ![Account and sign-in problems in Foxora v5.0.3](/docs/images/v5.0.3/auth/sign-in-native.webp) ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check connectivity, browser domain, account identity, clock, authorized devices, and subscription state. 4. Repeat the browser handoff and capture the exact failure. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/session-and-project-problems title: Task and project problems --- # Task and project problems Confirm the active local task and project path, folder availability, repository state, permissions, and free disk space; test read access before edits. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Confirm the active local task and project path, folder availability, repository state, permissions, and free disk space. 4. Test read access before edits. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/agent-crew-and-tool-problems title: Agent, crew, and tool problems --- # Agent, crew, and tool problems Check the selected configuration, instructions, model access, enabled skills and tools, approval state, and first failing call; test one agent and one tool in isolation. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check the selected configuration, instructions, model access, enabled skills and tools, approval state, and first failing call. 4. Test one agent and one tool in isolation. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/browser-and-computer-use-problems title: Browser and computer-use problems --- # Browser and computer-use problems Check the selected mode, bridge or OS permission, active profile and window, screen visibility, focus, and connection state; retest on a harmless target. ![Browser and computer-use problems in Foxora v5.0.3](/docs/images/v5.0.3/settings/permissions.webp) ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check the selected mode, bridge or OS permission, active profile and window, screen visibility, focus, and connection state; retest on a harmless target. 4. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 5. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/workflow-and-scheduled-task-problems title: Workflow and scheduled-task problems --- # Workflow and scheduled-task problems Check enabled state, trigger or next-run time, timezone, input mapping, selected agent, connections, approval wait, and first failed step; use a manual test run. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check enabled state, trigger or next-run time, timezone, input mapping, selected agent, connections, approval wait, and first failed step. 4. Use a manual test run. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/plugin-channel-and-mcp-problems title: Plugin, channel, and MCP problems --- # Plugin, channel, and MCP problems Check provider authorization, Composio connection or MCP transport, scopes, destination mapping, capability discovery, and latest error; reconnect only the failing layer. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Check provider authorization, Composio connection or MCP transport, scopes, destination mapping, capability discovery, and latest error. 4. Reconnect only the failing layer. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/collecting-diagnostics title: Collecting diagnostics --- # Collecting diagnostics Record versions, platform, time, reproduction steps, run or task identifiers, exact error, redacted logs, and expected versus actual behavior without exposing secrets. ![Collecting diagnostics in Foxora v5.0.3](/docs/images/v5.0.3/settings/system.webp) ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Reproduce the problem once with the smallest safe input and record the exact message, action, and time. 3. Record versions, platform, time, reproduction steps. 4. Run or task identifiers, exact error, redacted logs, and expected versus actual behavior without exposing secrets. 5. Change one reversible condition, then repeat the same test. Preserve logs or exports before any reset. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/troubleshooting/contacting-support title: Contacting support --- # Contacting support Search relevant documentation, prepare the redacted diagnostic bundle, describe impact and the smallest reproduction, include identifiers and time, then use the official support channel. ## Where this lives Open the failing surface and preserve the exact error before changing configuration. Troubleshooting starts by identifying the failing layer: installation, runtime, account, local project, agent configuration, tool, browser, automation, or external connection. > **Safety:** Redact tokens, passwords, private prompts, personal data, local paths, and proprietary file contents before sharing diagnostics. Avoid resets until evidence and local exports are preserved. ## Steps 1. Open the failing surface and preserve the exact error before changing configuration. 2. Search relevant documentation. 3. Prepare the redacted diagnostic bundle. 4. Describe impact and the smallest reproduction, include identifiers and time. 5. Use the official support channel. 6. **Confirm the result.** Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. ## Choose the right path - Use `foxora doctor` for CLI, engine, runtime, and authentication checks. - Use run or task logs for agent and automation failures. - Use connection tests for plugins, channels, MCP, browser, and providers. ## Confirm it worked - Repeat the identical minimal test, confirm the error is gone and the original workflow works, and remove any temporary diagnostic access or configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/overview title: Overview --- # Overview Find the correct source of truth for common questions and verify the answer against the current device, account, configuration, or external connection. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Find the correct source of truth for common questions and verify the answer against the current device, account, configuration, or external connection. 3. Identify whether the question concerns local data, account services, models and credits, agents and automation, or a connected external system before changing anything. 4. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/general-questions title: General questions --- # General questions Identify the Foxora surface involved, open its overview and current settings, confirm the installed version, and use the relevant step-by-step guide for the task. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Identify the Foxora surface involved. 3. Open its overview and current settings. 4. Confirm the installed version. 5. Use the relevant step-by-step guide for the task. 6. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/account-and-billing-questions title: Account and billing questions --- # Account and billing questions Open Account, Billing, and Usage, verify identity and plan, compare the recorded renewal or credit window, and use the secure billing portal for changes. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Open Account, Billing, and Usage. 3. Verify identity and plan. 4. Compare the recorded renewal or credit window. 5. Use the secure billing portal for changes. 6. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/models-and-credits-questions title: Models and credits questions --- # Models and credits questions Check whether the agent uses Foxora-managed access or BYOK, identify the model or processing tier, inspect Usage, and compare provider billing separately for BYOK. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Check whether the agent uses Foxora-managed access or BYOK. 3. Identify the model or processing tier. 4. Inspect Usage. 5. Compare provider billing separately for BYOK. 6. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/data-and-privacy-questions title: Data and privacy questions --- # Data and privacy questions Separate account data from local tasks and Memory Den, inspect device storage and connected-service permissions, and export or delete only the intended category. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Separate account data from local tasks and Memory Den. 3. Inspect device storage and connected-service permissions. 4. Export or delete only the intended category. 5. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/faq/agents-and-automation-questions title: Agents and automation questions --- # Agents and automation questions Choose agent, crew, bot, scheduled task, or workflow by the desired trigger and coordination pattern, then test the configuration with a bounded input. ## Where this lives Find the product area named in the question. The FAQ is a decision guide for common product questions. Each answer points to the setting or docs area where you can confirm the behavior in your own account and installation. > **Safety:** Do not treat a general answer as authorization for a destructive or externally visible action. Review the target and current product state before proceeding. ## Steps 1. Find the product area named in the question. 2. Choose agent, crew, bot, scheduled task, or workflow by the desired trigger and coordination pattern. 3. Test the configuration with a bounded input. 4. **Confirm the result.** Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. ## Choose the right path - Use Troubleshooting for a reproducible failure. - Use topic guides for a configuration procedure. - Use the API Reference or CLI command reference for exact programmatic syntax. ## Confirm it worked - Open the linked surface, confirm the effective account, device, agent, or connection, and use a small test to verify the answer for the current configuration. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/release-notes/overview title: Overview --- # Overview Use release notes to plan upgrades, identify client-specific versions, understand behavioral changes, and validate the installation afterward. ## Where this lives Open **Settings → About** to compare the installed version with the release notes. Release notes explain what changed, which client or runtime version is affected, required migrations, security fixes, known limitations, and deprecated behavior. > **Safety:** Install only signed official releases. Do not infer that the Desktop and CLI share a version number, and do not clear local data merely to resolve an update problem. ## Steps 1. Open **Settings → About** to compare the installed version with the release notes. 2. Use release notes to plan upgrades. 3. Identify client-specific versions. 4. Understand behavioral changes. 5. Validate the installation afterward. 6. Check the installed Desktop and CLI versions separately, read every release between the current and target versions, save active work, and preserve local-data exports before major upgrades. 7. **Confirm the result.** Restart the updated client and runtime, confirm the displayed versions, run diagnostics, and complete one small task, tool, and connection test relevant to your workflow. ## Choose the right path - Read Latest release for the newest published change. - Read V5 release notes when moving to the current product generation. - Read Known limitations and Deprecated features before relying on older behavior. ## Confirm it worked - Restart the updated client and runtime, confirm the displayed versions, run diagnostics, and complete one small task, tool, and connection test relevant to your workflow. - The screen, command, file, run, or destination named in this guide reflects the expected state. --- source: https://docs.foxora.ai/release-notes/latest-release title: Latest release --- # Foxora 5.0.4 Foxora 5.0.4 is the current desktop generation. The desktop release was published on August 5, 2026. It is the reliability release: the autonomous crew got a spine, and you got a choice about when to use it. ## Current versions Surfaces ship on their own cadence, so these numbers are deliberately not all the same. | Surface | Version | Availability | | --- | --- | --- | | Desktop | 5.0.4 | Universal and Apple Silicon macOS DMGs, Windows x64 installer | | CLI | 5.0.3 | macOS, Linux (gnu and musl), and Windows x64 | | Runtime | 5.0.0 | Provisioned by the installers | | Web app | Continuous | `https://app.foxora.ai` | | Cloud engine | Continuous | `https://engine.foxora.ai` | | Daytona workspace image | `foxora-base-v2` | New and lifecycle-recreated cloud workspaces | ## New - **Crew or Single model — your call, per task.** The composer's picker is a two-lane switch. **Crew** runs the autonomous multi-agent engine, which plans, delegates to specialists, and verifies its own work. **Single model** runs one model from your connected providers with no crew, no delegation, and no credits — you pay your provider directly. - **Agent notifications that hold.** When a long-running agent finishes or needs you, it lands in the Notification Center, durable across restarts and synced across your devices. Urgent ones also raise an OS banner. - **Source Control follows the active project.** The panel resolves the active project's working directory, including isolated worktrees, instead of a stale workspace root. - **The System tab tells the whole truth**: the live engine release, every component version, and connection tiles that say "Unreachable" only when something is genuinely unreachable. ## Improved - Crew runs route work to the right specialists — visual work goes through the designer with a rendered-output review, and the completion judge holds the quality line. - Steering mid-run is immediate and survives reloads. Damaged sessions heal their history instead of losing your prompt. - Sidebar run indicators are steady across the fleet rather than flapping between running and idle. - The Notification Center coalesces repeats into one truthful card per event. - Memory stops double-writing: one durable fact per insight, and old duplicates sweep themselves out as you work. ## Fixed - **Task history stays put.** Assistant replies and tool activity are written durably every turn, so reopening a task shows the whole conversation rather than only the prompts you typed. Writes survive a device reconnect instead of being dropped while the socket is down. - **Voice calls are calls again.** The agent answers out loud and keeps the conversation; only real work is handed to the crew, and the finished answer is spoken back. Speaking no longer files your words as a task. - **Memory search works on desktop.** Embeddings are stored and ranked locally again. - Session task-state errors no longer masquerade as connection problems. - Notification taps navigate, and never silently approve anything. - Auto mode decides, states the choice, and moves on instead of interrogating you. - A task's selected crew or single model sticks instead of reverting to the default crew on the next poll. ## Known limitation With the cloud engine — the desktop default — git runs where the engine runs, so a local repository reads as "not a git repository". Source Control is accurate today only with a local engine. Routing git to your machine is the next release. ## Download - [Download Foxora 5.0.4](https://foxora.ai/download) - [Read the signed release on GitHub](https://github.com/foxora-ai/releases/releases/tag/desktop-v5.0.4) macOS ships two notarized images: `Foxora_5.0.4_universal.dmg` runs on Apple Silicon and Intel, and `Foxora_5.0.4_AppleSilicon.dmg` is about half the size but Apple Silicon only. Windows users download `Foxora_5.0.4_x64-setup.exe`; SmartScreen may still show an unrecognized-publisher warning while publisher reputation is established. Already on macOS? Install or update in place with one command: ```bash curl -fsSL https://foxora.ai/install.sh | sh -s -- --update ``` It verifies the download's checksum and signature, and refuses to install anything that does not match. ## Cloud workspace transition Foxora does not delete active user workspaces merely to replace their base image. Existing Daytona v1 workspaces remain available and move to v2 when the normal backup-and-recreation lifecycle applies. New or recreated workspaces use v2 immediately. ## Upgrade verification 1. Open **Settings → About** or run `foxora --version`. 2. Install the signed update and restart the client. 3. Confirm the displayed version is 5.0.4. 4. Complete a small session and workspace task. 5. If you use cloud execution, confirm the workspace starts or resumes and a simple command succeeds. No manual data migration is required for a normal 5.0.3-to-5.0.4 desktop update. --- source: https://docs.foxora.ai/release-notes/v5-release-notes title: V5 release notes --- # Foxora v5 release notes Foxora v5 unifies the desktop, web, CLI, shared UI, local runtime, cloud engine, and Memory Den around the same task and tool contracts. ![V5 release notes in Foxora v5.0.3](/docs/images/v5.0.3/settings/about.webp) ## Workspace and tasks - Tasks are durable work threads with their own transcript, project context, agent or crew selection, autonomy mode, and Stage state. - Projects provide the filesystem boundary; tasks provide conversational and execution history. - Forking, recovery, checkpoints, archiving, and run history keep long work inspectable and recoverable. - Desktop and CLI can use the same local runtime on one device. The web runtime uses cloud execution and does not automatically copy device-local tasks or files. ## Foxora workspace and Studio - The **foxora** workspace contains New Task, Projects, Workflows, Schedules, the project list, and the task list. - **studio** contains Bots, Agents, Crews, Skills, Tools, Memory, and Plugins. - **Runs**, transcript view, and right-panel controls live in the active task header. - Account, Appearance, Voice, Models, Browser, Plugins, Channels, Brand, MCP Server, Permissions, Data, System, Keyboard, and About live in Settings. An **Extensions** Studio surface is not part of v5.0.3. Use Skills for procedural knowledge, Plugins for supported application accounts, MCP Server for exposed Foxora capabilities, and the browser extension only for authorized browser-session access. ## Autonomy and approvals Each task can use Ask Permissions, Accept Edits, Auto, or Bypass Permissions. The mode controls approval pauses; it does not make publishing, messaging, purchasing, deletion, or production changes reversible. ## Desktop and CLI - Desktop 5.0.3 ships a signed, notarized universal macOS package and a Windows x64 installer. - CLI 5.0.3 provides the interactive terminal, one-shot `run`, account commands, usage, models, session and crew lists, setup, update, and diagnostics. - Signed updater manifests let compatible desktop installations discover newer releases. ## Web and cloud runtime - The web app uses the same shared Foxora interface with a cloud-owned runtime boundary. - Daytona workspaces provide isolated project execution and resume after provider auto-stop. - The v2 workspace image adds a verified browser and document toolchain. - Existing v1 workspaces are preserved until the normal lifecycle recreates them. ## Compatibility Normal upgrades preserve application data. Before major troubleshooting, export anything important and distinguish between account data, device-local tasks, Memory Den, project files, and cloud workspace data. See **Latest release** for the exact current versions and **Known limitations** before relying on platform-specific functionality. --- source: https://docs.foxora.ai/release-notes/previous-releases title: Previous releases --- # Previous releases Use the installed version under **Settings → About** or `foxora --version`, then read every release between that version and the target. ## Desktop 5.0.3 — July 27, 2026 - Refreshed credentials earlier in long-running sessions and recovered more reliably from interrupted streams. - Opened the produced application surface from preview routing instead of a blank or unrelated preview. - Kept project-aware generation attached to the intended workspace. - Hardened managed browser packaging for the notarized macOS application. - Resumed stopped Daytona sandboxes and refreshed stale handles before executing work. [View Desktop 5.0.3](https://github.com/foxora-ai/releases/releases/tag/desktop-v5.0.3) ## Desktop 5.0.2 — July 17, 2026 - Added a Windows 10/11 installer containing Desktop, Core, CLI, browser runtime, and prerequisites. - Preserved sessions, settings, login, Memory Den, and workspaces during upgrades. - Improved session loading, WebView memory behavior, device relay, and background task continuity. - Published a universal Apple Silicon and Intel macOS build. [View Desktop 5.0.2](https://github.com/foxora-ai/releases/releases/tag/desktop-v5.0.2) ## Desktop 5.0.1 — July 10, 2026 - Registered Foxora Core as an app-owned macOS background service. - Verified the secure engine relay, local workspace, and Memory Den. - Published signed, notarized, stapled macOS app and DMG artifacts. - Added a signed in-app updater bundle. [View Desktop 5.0.1](https://github.com/foxora-ai/releases/releases/tag/desktop-v5.0.1) ## Desktop 5.0.0 — June 21, 2026 - Introduced the v5 local engine and consolidated local state under the Foxora data directory. - Prevented inline reasoning tags from leaking into the visible conversation. - Allowed answered approval questions to resume the active turn. [View Desktop 5.0.0](https://github.com/foxora-ai/releases/releases/tag/desktop-v5.0.0) ## Before upgrading from an older release 1. Finish active runs and save project work. 2. Export important local sessions or data where required. 3. Read **Known limitations** and **Deprecated features**. 4. Install the latest signed release. 5. Verify the version, local runtime, authentication, and one representative task. Do not delete local data as an upgrade step unless the specific release instructions require it. --- source: https://docs.foxora.ai/release-notes/known-limitations title: Known limitations --- # Known limitations These limitations apply to Foxora 5.0.3 unless a newer release note says otherwise. ## Platform availability - A Linux desktop package is not currently published. Use the Linux CLI or web app. - The Windows desktop build is x64. Windows on ARM is not a published desktop target. - Windows SmartScreen may display an unrecognized-publisher warning while code-signing reputation is established. ## Local and cloud data - Signing into the same account on another computer does not automatically copy device-local sessions, checkpoints, or Memory Den data. - The web app does not automatically receive folders from your computer. Clone, upload, or create the project in the cloud workspace as appropriate. - Existing Daytona v1 workspaces are preserved and transition to the v2 image through the normal backup-and-recreation lifecycle; they are not destructively replaced during the release. ## API access The v5.0.3 gateway accepts signed-in account access tokens for authenticated user routes. Self-service long-lived developer API keys are not exposed in the current product. Do not copy internal service credentials from application storage or embed account tokens in public clients. ## Permissions and connected services - Browser and Computer Use behavior depends on operating-system permissions and the active browser mode. - A plugin showing connected does not by itself prove that every tool, trigger, channel, or outbound reply is configured. Test the exact action and destination. - External provider limits and BYOK charges remain separate from Foxora credits. ## Reporting a problem Record the Foxora version, operating system, time, session or run identifier, exact error, and the smallest safe reproduction. Redact access tokens, provider keys, personal content, and unrelated project files before sharing logs. --- source: https://docs.foxora.ai/release-notes/deprecated-features title: Deprecated features --- # Deprecated and renamed features Foxora v5 consolidates overlapping surfaces so documentation and the application use one name for each capability. ## Connections became Plugins The standalone **Connections** settings surface is retired. Use **Settings → Plugins** for supported application accounts and **Settings → MCP Server** for custom tool servers. Older instructions that say “open Connections” should be translated according to the integration: | Older term | Current destination | | --- | --- | | Application connection | Settings → Plugins | | Custom MCP connection | Settings → MCP Server | | Messaging integration | Settings → Channels | | Model-provider key | Models | ## Extensions Studio surface An **Extensions** section is not part of the v5.0.3 Studio. Choose the supported replacement: - Skill — reusable instructions, scripts, references, and assets. - Plugin — supported third-party application authorization and tools. - MCP server — custom tools and resources. - Browser extension — access to an explicitly authorized existing browser session. ## Older installer names Installer filenames carry the release, so any instruction naming a fixed filename goes stale. Two names in particular are retired: - `Foxora-Setup.exe` — the unversioned Windows installer, last published for 5.0.3. Windows now downloads `Foxora__x64-setup.exe`. - `Foxora__aarch64.dmg` — the old Apple Silicon image name. Since 5.0.4 it is `Foxora__AppleSilicon.dmg`. Architecture-specific macOS downloads are not deprecated: 5.0.4 publishes both a universal image and a smaller Apple Silicon one. Take the universal image when scripting or when unsure. ## Older CLI syntax Use plural list commands: `foxora sessions`, `foxora crews`, and `foxora agents`. Commands such as `foxora crew`, `foxora queue`, `foxora recall`, and `foxora org` are not part of CLI v5.0.3. Before removing an older workflow or script, replace its command or destination and test the updated path with a low-risk input.