PluginBench
MCP Server
Active
MIT

Browsentic MCP Server

io.github.imshaikot/browsentic

Give your AI agent control of your real, logged-in browser via a side panel or MCP.

RepositoryRegistry

What is the Browsentic MCP server?

Browsentic is a browser extension that lets AI agents control your actual logged-in browser through a side panel or MCP server. It works with Claude, Cursor, and other agent CLIs, providing 50+ deterministic tools for navigation, form filling, file handling, and more—all locally, with no cloud or authentication required.

Browsentic bridges your real browser and AI agents by running as a Manifest V3 extension in Chrome, Firefox, Edge, Brave, and Arc. Instead of headless automation, agents interact with your already-logged-in sessions, read and upload files, solve captchas, and perform complex workflows. It's local-first, privacy-focused, and works with every major agent CLI or as an optional MCP server.

How to install Browsentic

Copy-paste configuration for popular MCP clients.

transport: stdio
Config generated by PluginBench — verify against the source before use.
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "browsentic": {
      "command": "npx",
      "args": [
        "-y",
        "browsentic",
        "mcp"
      ]
    }
  }
}

Tools & capabilities

Tools this server exposes to the agent.

  • Page navigation — Full browser navigation including go back, forward, reload, and URL navigation
  • Element interaction — Click, type, focus, and interact with page elements
  • Form filling — Fill and submit forms with agent-provided data
  • File operations — Read files, upload to pages, and download from pages
  • Page inspection — Get page content, element details, and visual information
  • Captcha solving — Automatically solve captchas encountered during browsing
  • Custom tool creation — Write and run live tools for tasks not covered by built-in tools
  • Recording and playback — Record actions via voice, text, or demonstration for agent learning
  • Scheduling and monitoring — Monitor pages and schedule recurring tasks
  • Site mapping — Create custom skills and site-specific automation patterns

Use cases

  • Automate multi-step workflows on websites you're already logged into without headless browser setup
  • Fill out forms, upload documents, and download results while maintaining your active sessions
  • Monitor web pages for changes and trigger actions on a schedule
  • Teach an agent how to use a custom or niche web application by showing it once
  • Combine browser automation with file handling to process batches of documents through web services

Browsentic MCP server FAQ

What is Browsentic?

Browsentic is a browser extension that gives AI agents (Claude, Cursor, etc.) control of your real, logged-in browser. It provides 50+ tools for navigation, form filling, file handling, and more—all running locally with no cloud or authentication.

Is Browsentic free?

Yes, Browsentic is MIT licensed and free. It's open-source, local-first, with no cloud component, telemetry, or account required.

How do I install it in Cursor or Claude?

Install via npm (npx browsentic setup) or use the macOS app (curl -fsSL https://browsentic.com/install.sh | sh). Then pair it with your agent CLI. For MCP clients like Claude Code or Cursor, enable the optional MCP server mode.

Does it require authentication or API keys?

No. Browsentic uses your browser's existing logged-in sessions. Everything runs locally on 127.0.0.1 with no external services, cloud accounts, or API keys needed.

Which browsers does it support?

Browsentic works in Chrome, Firefox, Edge, Brave, and Arc. It's built as a Manifest V3 extension and also runs as a native macOS and Windows app.

Can I use it with other MCP clients?

Yes. Browsentic can run as an optional MCP server, so any MCP-compatible client (Claude Code, Cursor, Zed, etc.) can drive the same browser instance.

README (reference)

Source of truth, from the repository.

https://github.com/user-attachments/assets/e9f146d3-cb14-416e-b8d7-68c4e81cbd0d

<div align="center">

Browsentic – a completely integrated agentic browser extension

<p> <a href="https://www.npmjs.com/package/browsentic"><img src="https://img.shields.io/npm/v/browsentic?logo=npm" alt="npm version"></a> <a href="https://www.npmjs.com/package/browsentic"><img src="https://img.shields.io/npm/dm/browsentic?logo=npm" alt="npm downloads"></a> <a href="https://github.com/imshaikot/browsentic/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/imshaikot/browsentic/ci.yml?branch=main&label=CI&logo=githubactions&logoColor=white" alt="CI"></a> <a href="https://nodejs.org"><img src="https://img.shields.io/node/v/browsentic?logo=nodedotjs&logoColor=white" alt="node"></a> <a href="LICENSE"><img src="https://img.shields.io/npm/l/browsentic" alt="license"></a> </p> <p> <a href="docs/guide/install.md"><img src="https://img.shields.io/badge/manifest-V3-4285F4?logo=googlechrome&logoColor=white" alt="Manifest V3"></a> <a href="https://github.com/imshaikot/browsentic/releases/latest"><img src="https://img.shields.io/github/actions/workflow/status/imshaikot/browsentic/release.yml?label=macOS%20build&logo=apple" alt="macOS build"></a> <a href="https://github.com/imshaikot/browsentic/releases/latest"><img src="https://img.shields.io/github/actions/workflow/status/imshaikot/browsentic/release.yml?label=Windows%20build&logo=data:image/svg%2Bxml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAyNCAyNCI+PHBhdGggZmlsbD0iI2ZmZiIgZD0iTTAgMy40IDkuOCAydjkuNUgwek0xMSAxLjggMjQgMHYxMS41SDExek0wIDEyLjVoOS44VjIyTDAgMjAuNnpNMTEgMTIuNWgxM1YyNGwtMTMtMS44eiIvPjwvc3ZnPg==" alt="Windows build"></a> <a href="https://github.com/imshaikot/browsentic/releases/latest"><img src="https://img.shields.io/github/actions/workflow/status/imshaikot/browsentic/release.yml?label=Firefox%20build&logo=firefoxbrowser&logoColor=white" alt="Firefox build"></a> </p>

No account · No auth · No API key · No cloud · No telemetry · Local first

Your real, logged-in browser, driven from an AI side panel by the agent CLI you already have.<br> Also an optional MCP server, so any MCP client can drive the same browser.

<p> <strong>Works in</strong><br> <a href="docs/guide/install.md"><img src="https://browsentic.com/icons/chrome.svg" width="40" height="40" alt="Chrome" title="Chrome"></a>&emsp; <a href="docs/guide/install.md#firefox"><img src="https://browsentic.com/icons/firefox.svg" width="40" height="40" alt="Firefox" title="Firefox"></a>&emsp; <a href="docs/guide/install.md"><img src="https://browsentic.com/icons/edge.svg" width="40" height="40" alt="Edge" title="Edge"></a>&emsp; <a href="docs/guide/install.md"><img src="https://browsentic.com/icons/brave.svg" width="40" height="40" alt="Brave" title="Brave"></a>&emsp; <a href="docs/guide/install.md"><img src="https://browsentic.com/icons/arc.svg" width="40" height="40" alt="Arc" title="Arc"></a> </p> <p> <strong>Runs on</strong><br> <a href="https://claude.com/claude-code"><img src="https://browsentic.com/icons/claude.svg" width="40" height="40" alt="Claude Code" title="Claude Code"></a>&emsp; <a href="https://developers.openai.com/codex/cli"><picture><source media="(prefers-color-scheme: dark)" srcset="https://browsentic.com/icons/openai-dark.svg"><img src="https://browsentic.com/icons/openai.svg" width="40" height="40" alt="Codex" title="Codex"></picture></a>&emsp; <a href="https://antigravity.google/docs/cli/install"><img src="https://browsentic.com/icons/antigravity.svg" width="40" height="40" alt="Antigravity" title="Antigravity"></a>&emsp; <a href="https://github.com/mistralai/mistral-vibe"><img src="https://browsentic.com/icons/mistral.svg" width="40" height="40" alt="Mistral Vibe (beta)" title="Mistral Vibe (beta)"></a>&emsp; <a href="https://docs.x.ai/build/overview"><picture><source media="(prefers-color-scheme: dark)" srcset="https://browsentic.com/icons/xai-dark.svg"><img src="https://browsentic.com/icons/xai.svg" width="40" height="40" alt="Grok Build (beta)" title="Grok Build (beta)"></picture></a>&emsp; <a href="https://cursor.com/docs/cli/overview"><picture><source media="(prefers-color-scheme: dark)" srcset="https://browsentic.com/icons/cursor-dark.svg"><img src="https://browsentic.com/icons/cursor.svg" width="40" height="40" alt="Cursor CLI (beta)" title="Cursor CLI (beta)"></picture></a>&emsp; <a href="https://qwenlm.github.io/qwen-code-docs/en/"><picture><source media="(prefers-color-scheme: dark)" srcset="https://browsentic.com/icons/qwen-dark.svg"><img src="https://browsentic.com/icons/qwen.svg" width="40" height="40" alt="Qwen Code (beta)" title="Qwen Code (beta)"></picture></a>&emsp; <a href="https://opencode.ai/docs/"><picture><source media="(prefers-color-scheme: dark)" srcset="https://browsentic.com/icons/opencode-dark.svg"><img src="https://browsentic.com/icons/opencode.svg" width="40" height="40" alt="OpenCode (beta)" title="OpenCode (beta)"></picture></a> </p> </div>

Quick Start

Any platform — with Node.js 20 or newer (Windows is experimental):

npx browsentic setup

macOS — one line installs Browsentic.app, which brings Node and everything else:

curl -fsSL https://browsentic.com/install.sh | sh

Windows — one line in PowerShell installs the Windows app, which brings Node and everything else (experimental):

irm https://browsentic.com/install.ps1 | iex

Full install and setup guide, Firefox included: browsentic.com/docs/guide/install

Key Capabilities

<table> <tr> <td>🧭 <a href="docs/guide/features/conversations.md"><b>Side panel, human in the loop</b></a></td> <td>🤖 <a href="docs/guide/agents.md"><b>Every major agent CLI</b></a></td> <td>🎯 <a href="docs/guide/features/page-actions.md"><b>50+ deterministic tools</b></a></td> </tr> <tr> <td>🌐 <a href="docs/guide/install.md"><b>All Chromium + Firefox</b></a></td> <td>🗺️ <a href="docs/guide/features/page-actions.md#moving-around"><b>Full browser navigation</b></a></td> <td>🧩 <a href="docs/guide/features/captcha.md"><b>Solves captchas on the way</b></a></td> </tr> <tr> <td>📄 <a href="docs/guide/features/files.md#files-you-hand-it"><b>Reads files, acts on them</b></a></td> <td>↕️ <a href="docs/guide/features/files.md#download-here-upload-there"><b>Uploads & downloads</b></a></td> <td>🧠 <a href="docs/guide/features/site-maps.md"><b>Creates its own skills</b></a></td> </tr> <tr> <td>⚡ <a href="docs/guide/features/page-actions.md#repeating-a-job-and-doing-what-no-tool-covers"><b>Writes & runs live tools</b></a></td> <td>🎙️ <a href="docs/guide/features/recordings.md"><b>Voice, text or show it once</b></a></td> <td>👁️ <a href="docs/guide/features/a-eye.md"><b>Point at what you mean</b></a></td> </tr> <tr> <td>⏱️ <a href="docs/guide/features/scheduling.md"><b>Monitors & schedules</b></a></td> <td>🛡️ <a href="docs/guide/approvals.md"><b>Guardrails & approvals</b></a></td> <td>🔌 <a href="docs/guide/mcp-clients.md"><b>Optional MCP server</b></a></td> </tr> </table>

How It Works

How an instruction becomes a click

<details> <summary>The same flow, as text</summary> 
You ──speak or type──> Extension ──local WebSocket──> Daemon ──spawns──> your agent CLI
                            ▲                              (claude │ codex │ agy │ vibe │ grok)
                            └──────────────── page actions ─────────────────────┘

Any MCP client ──stdio──> browsentic mcp ──> the same daemon ──> the same browser
</details>

The extension dials out to the daemon, because a Manifest V3 service worker cannot listen for connections. One daemon owns the browser link, so several MCP clients can share one browser. Everything binds to 127.0.0.1.

Resources

Privacy and Security

Nothing connects until you pair, both ends prove themselves, consequential actions ask first, and credentials on a page are sealed before the agent sees them. The full model is in SECURITY.md, and what it does not cover is in Limits.

Contributing

Bugs and ideas are welcome — start at CONTRIBUTING.md.

License

Browsentic is MIT licensed.

  • Source Available: Always visible source code
  • Local First: No cloud component, no telemetry, no account
  • Extensible: Add your own page capabilities, skills and agent runners

What does Browsentic mean?

Short answer: "Browse" + "agentic".

Long answer: Most browser automation asks you to hand the work to a different browser — a headless one, in a container, logged in to nothing. Browsentic is the other way round: the agentic part happens in the browser you are already looking at, with the sessions you are already signed in to. The name is the thesis — browsing, made agentic, where you already browse.

Related MCP servers

ANAnneal logo

Anneal

Active

Zero-friction personal operating system: habits, tasks, and FSRS-5 study retention via remote MCP.

0
MIT
View repository →
IOio.github.imviky-ctrl/tickerr logo

io.github.imviky-ctrl/tickerr

Maintained

Live status, API pricing and rate limits for ChatGPT, Claude, Gemini, Cursor and 42+ AI tools.

1
JavaScript
View repository →
IOio.github.imysfylmz/xrechnung-mcp logo

io.github.imysfylmz/xrechnung-mcp

Validate and parse German e-invoices (XRechnung, ZUGFeRD, Factur-X) offline. No API key.

View repository →
IOio.github.inSideos-designs/cowork-qa logo

io.github.inSideos-designs/cowork-qa

Maintained

Goal-driven Playwright browser sessions for LLMs, with full action traces and aria-snapshots.

0
JavaScript
MIT
View repository →
IOio.github.inamprotocol/inam-mcp logo

io.github.inamprotocol/inam-mcp

Active

Check an agent's INAM reputation before trusting it, and sign a receipt when work is done.

0
TypeScript
Apache-2.0
View repository →
INiNavi Maps Docs logo

iNavi Maps Docs

Active

iNavi Maps docs for AI (South Korea): API specs, Web JS SDK reference, HTML map templates

0
TypeScript
MIT
View repository →