* fix: remove filesystem tools when no workspace is selected - Make workingDir optional on ResolvedAgentConfig - Remove resolveSessionDir() fallback that always created a session dir, masking the no-workspace state and keeping filesystem tools available - Gate buildFilesystemToolSet() on workingDir being defined - Add workspace change detection mid-conversation — rebuilds the agent session when workspace is added, removed, or switched (same pattern as existing MCP server change detection) - download_file falls back to tmpdir() when no workspace is set - Memory/soul tools are unaffected — they use ~/BrowserOS/ paths * fix: sanitize message history when session rebuilds with different tools When a session is rebuilt due to workspace or MCP changes, the carried-over message history may contain tool parts for tools that no longer exist in the new session. The AI SDK validates messages against the current toolset and rejects parts with no matching schema. - Add toolNames getter to AiSdkAgent exposing registered tool names - Add sanitizeMessagesForToolset() to strip tool parts referencing removed tools from carried-over messages - Apply sanitization in both MCP and workspace session rebuilds * fix: prepend tool-change context to user message on session rebuild When workspace or MCP integrations change mid-conversation, prepend a [Context: ...] block to the user's message explaining what changed. This prevents the LLM from hallucinating tool usage based on patterns in the carried-over conversation history. Context messages vary by change type: - Workspace removed: lists unavailable filesystem tools, suggests selecting a working directory - Workspace added: confirms filesystem tools are available with path - Workspace switched: notes the new working directory - MCP changed: notes that some integration tools may have changed Only fires on the first message after a rebuild. Invisible in the UI. * fix: make MCP change context specific about which apps were added/removed Diff the old and new MCP server keys to produce specific context like: - "The following app integrations were disconnected: Gmail, Slack." - "The following app integrations were connected: Linear." instead of a generic "some tools may no longer be available" message. * refactor: extract shared rebuildSession helper in ChatService Eliminates the duplicated 20-line dispose→create→sanitize→store flow that existed separately in both the MCP and workspace change-detection blocks. Co-authored-by: Dani Akash <DaniAkash@users.noreply.github.com> * test: add sanitizeMessagesForToolset test suite Tests for the message sanitization that runs when a session rebuilds with a different toolset (workspace or MCP change mid-conversation): - Preserves messages with no tool parts - Preserves tool parts when tool is in the toolset - Strips tool parts when tool is NOT in the toolset - Strips multiple removed tool parts from same message - Keeps browser tools while removing filesystem tools - Removes messages that become empty after stripping - Preserves non-tool parts (reasoning, step-start, file) - Returns same references when no filtering needed - Handles empty message array and empty toolset * style: fix biome formatting in chat-service.ts --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
BrowserOS is an open-source Chromium fork that runs AI agents natively. The privacy-first alternative to ChatGPT Atlas, Perplexity Comet, and Dia.
Use your own API keys or run local models with Ollama. Your data never leaves your machine.
Documentation · Discord · Slack · Twitter · Feature Requests
Quick Start
- Download and install BrowserOS — macOS · Windows · Linux (AppImage) · Linux (Debian)
- Import your Chrome data (optional) — bookmarks, passwords, extensions all carry over
- Connect your AI provider — Claude, OpenAI, Gemini, ChatGPT Pro via OAuth, or local models via Ollama/LM Studio
Features
| Feature | Description | Docs |
|---|---|---|
| AI Agent | 53+ browser automation tools — navigate, click, type, extract data, all with natural language | Guide |
| MCP Server | Control the browser from Claude Code, Gemini CLI, or any MCP client | Setup |
| Workflows | Build repeatable browser automations with a visual graph builder | Docs |
| Cowork | Combine browser automation with local file operations — research the web, save reports to your folder | Docs |
| Scheduled Tasks | Run agents on autopilot — daily, hourly, or every few minutes | Docs |
| Memory | Persistent memory across conversations — your assistant remembers context over time | Docs |
| SOUL.md | Define your AI's personality and instructions in a single markdown file | Docs |
| LLM Hub | Compare Claude, ChatGPT, and Gemini responses side-by-side on any page | Docs |
| 40+ App Integrations | Gmail, Slack, GitHub, Linear, Notion, Figma, Salesforce, and more via MCP | Docs |
| Vertical Tabs | Side-panel tab management — stay organized even with 100+ tabs open | Docs |
| Ad Blocking | uBlock Origin + Manifest V2 support — 10x more protection than Chrome | Docs |
| Cloud Sync | Sync browser config and agent history across devices | Docs |
| Skills | Custom instruction sets that shape how your AI assistant behaves | Docs |
| Smart Nudges | Contextual suggestions to connect apps and use features at the right moment | Docs |
Demos
BrowserOS agent in action
Install BrowserOS as MCP and control it from claude-code
https://github.com/user-attachments/assets/c725d6df-1a0d-40eb-a125-ea009bf664dc
Use BrowserOS to chat
https://github.com/user-attachments/assets/726803c5-8e36-420e-8694-c63a2607beca
Use BrowserOS to scrape data
https://github.com/user-attachments/assets/9f038216-bc24-4555-abf1-af2adcb7ebc0
Install browseros-cli
Use browseros-cli to launch and control BrowserOS from the terminal or from AI coding agents like Claude Code.
macOS / Linux:
curl -fsSL https://cdn.browseros.com/cli/install.sh | bash
Windows:
irm https://cdn.browseros.com/cli/install.ps1 | iex
After install, run browseros-cli init to connect the CLI to your running BrowserOS instance.
LLM Providers
BrowserOS works with any LLM. Bring your own keys, use OAuth, or run models locally.
| Provider | Type | Auth |
|---|---|---|
| Kimi K2.5 | Cloud (default) | Built-in |
| ChatGPT Pro/Plus | Cloud | OAuth |
| GitHub Copilot | Cloud | OAuth |
| Qwen Code | Cloud | OAuth |
| Claude (Anthropic) | Cloud | API key |
| GPT-4o / o3 (OpenAI) | Cloud | API key |
| Gemini (Google) | Cloud | API key |
| Azure OpenAI | Cloud | API key |
| AWS Bedrock | Cloud | IAM credentials |
| OpenRouter | Cloud | API key |
| Ollama | Local | Setup |
| LM Studio | Local | Setup |
How We Compare
| BrowserOS | Chrome | Brave | Dia | Comet | Atlas | |
|---|---|---|---|---|---|---|
| Open Source | ✅ | ❌ | ✅ | ❌ | ❌ | ❌ |
| AI Agent | ✅ | ❌ | ❌ | ❌ | ✅ | ✅ |
| MCP Server | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
| Visual Workflows | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
| Cowork (files + browser) | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
| Scheduled Tasks | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ |
| Bring Your Own Keys | ✅ | ❌ | ✅ | ❌ | ❌ | ❌ |
| Local Models (Ollama) | ✅ | ❌ | ✅ | ❌ | ❌ | ❌ |
| Local-first Privacy | ✅ | ❌ | ✅ | ❌ | ❌ | ❌ |
| Ad Blocking (MV2) | ✅ | ❌ | ✅ | ❌ | ✅ | ❌ |
Detailed comparisons:
- BrowserOS vs Chrome DevTools MCP — developer-focused comparison for browser automation
- BrowserOS vs Claude Cowork — getting real work done with AI
- BrowserOS vs OpenClaw — everyday AI assistance
Architecture
BrowserOS is a monorepo with two main subsystems: the browser (Chromium fork) and the agent platform (TypeScript/Go).
BrowserOS/
├── packages/browseros/ # Chromium fork + build system (Python)
│ ├── chromium_patches/ # Patches applied to Chromium source
│ ├── build/ # Build CLI and modules
│ └── resources/ # Icons, entitlements, signing
│
├── packages/browseros-agent/ # Agent platform (TypeScript/Go)
│ ├── apps/
│ │ ├── server/ # MCP server + AI agent loop (Bun)
│ │ ├── agent/ # Browser extension UI (WXT + React)
│ │ ├── cli/ # CLI tool (Go)
│ │ ├── eval/ # Benchmark framework
│ │ └── controller-ext/ # Chrome API bridge extension
│ │
│ └── packages/
│ ├── agent-sdk/ # Node.js SDK (npm: @browseros-ai/agent-sdk)
│ ├── cdp-protocol/ # CDP type bindings
│ └── shared/ # Shared constants
| Package | What it does |
|---|---|
packages/browseros |
Chromium fork — patches, build system, signing |
apps/server |
Bun server exposing 53+ MCP tools and running the AI agent loop |
apps/agent |
Browser extension — new tab, side panel chat, onboarding, settings |
apps/cli |
Go CLI — control BrowserOS from the terminal or AI coding agents |
apps/eval |
Benchmark framework — WebVoyager, Mind2Web evaluation |
agent-sdk |
Node.js SDK for browser automation with natural language |
cdp-protocol |
Type-safe Chrome DevTools Protocol bindings |
Contributing
We'd love your help making BrowserOS better! See our Contributing Guide for details.
Agent development (TypeScript/Go) — see the agent monorepo README for setup instructions.
Browser development (C++/Python) — requires ~100GB disk space. See packages/browseros for build instructions.
Credits
- ungoogled-chromium — BrowserOS uses some patches for enhanced privacy. Thanks to everyone behind this project!
- The Chromium Project — at the core of BrowserOS, making it possible to exist in the first place.
License
BrowserOS is open source under the AGPL-3.0 license.
Copyright © 2025 Felafax, Inc.
Stargazers
Thank you to all our supporters!
Built with ❤️ from San Francisco
