epiral/bb-browser

Your browser is the API. CLI + MCP server for AI agents to control Chrome with your login state.

What it solves

AI agents typically struggle to access the live web because most websites lack APIs or have strict anti-bot protections. bb-browser solves this by allowing agents to use a user's actual browser session—including login states and cookies—as the API, effectively turning the entire internet into a a machine-readable interface without needing API keys or scrapers.

How it works

Instead of using headless browsers or isolated environments, bb-browser connects to your real Chrome browser via a daemon and the Chrome DevTools Protocol (CDP). It executes JavaScript (eval) directly within your active browser tabs, performs authenticated fetch() calls using your existing cookies, or interacts with the page's own internal modules (like Webpack). This makes the agent's actions invisible to anti-bot systems because the requests originate from a legitimate, logged-in user session.

Who it’s for

  • AI Agent Developers: Those building agents that need to perform real-world tasks across multiple platforms (e.g., social media, finance, research).
  • MCP Users: Users of Claude Code or Cursor who want to give their AI tools the ability to browse the web authentically.
  • LlamaIndex/OpenClaw Users: Users who want integrated browser automation without needing separate extensions.

Highlights

  • Massive Platform Support: Includes 103 commands across 36 platforms including Twitter, Reddit, GitHub, YouTube, and LinkedIn.
  • MCP Integration: Native support for Model Context Protocol (MCP) for seamless use with AI IDEs.
  • Community-Driven Adapters: A system of lightweight JS adapters that can be quickly reverse-engineered and added by AI agents themselves.
  • Full Browser Automation: Beyond site-specific commands, it provides a CLI for snapshots, clicking, filling forms, and capturing network traffic.
  • passage-based JSON output: All commands support --json and --jq for structured data extraction.

Related

  • Project
    vibheksoni/stealth-browser-mcp

    An MCP-compatible server for stealthy browser automation that allows AI agents to bypass anti-bot checks and Cloudflare challenges using real Chrome instances.

  • Project
    actionbook/actionbook

    An AI browser agent that uses a Chrome extension and MCP connector to operate websites concurrently and access data behind logins to deliver finished work.

  • Project
    browser-use/browser-use

    A library and framework that enables AI agents to control a web browser like a human to automate tasks such as filling forms and extracting data.

  • Project
    unbrowse-ai/unbrowse

    Unbrowse is an open‑source tool that lets AI agents call a website’s own API endpoints instead of driving a full browser. It learns those endpoints by observing a real browser once, caches them, and then resolves future intents with a single HTTP request, falling back to browser automation when needed. The project includes a CLI, a TypeScript SDK, account/credit billing, and an optional Agent Skill for easy integration.

  • Project
    AgentDeskAI/browser-tools-mcp

    An MCP server and Chrome extension that gives AI agents access to a user's active browser session for real-time debugging, network inspection, and Lighthouse audits.