M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
B

BrowserOS MCP Server

Community
MCP server exposing Chromium browser automation capabilities through BrowserOS
How to install ↓ Compare GitHub source repository ↗
Category
Browser #30 of 32
Stars
★ 13.9k Very popular
Transport
stdio (local process)
Runtime
Prebuilt binary
Credentials
No credential needed
License
AGPL-3.0
Last commit
34FMRS · D

BrowserOS MCP Server is an official MCP server providing AI agents with local browser automation, reusing real login sessions, supporting parallel tasks and session replay. Compared to cloud browsers and headless drivers, it emphasizes privacy and user control. Suitable for developers handling real workflows, but requires caution regarding agent permissions and potential risks.

Strongest · Maintenance 10/20 Weakest · Reliability 4/20

Reliability
4/20
Security and permissions
5/20
Maintenance
10/20
Documentation
8/20
Setup experience
7/20
Why each score
Reliability 4/20
Only server.json and the README were available; there is no source, tool schema, tests, or CI to compare against. The manifest's repository field points to browseros/browseros-server while the real repo is browseros-ai/BrowserOS, so the metadata is self-contradictory. Whether the npm package [email protected] exists or whether the stdio handshake and tool behaviors align cannot be verified. A plausible happy path is shown only by README claims, with no execution evidence and a static cap of 12; score 4.
Security and permissions 5/20
No red-line issue such as malware, credential theft, or covert exfiltration appears in the supplied material. However, this MCP server is designed to drive a browser with real logged-in sessions, a high-privilege, high-data-exposure surface. No tool-level least privilege, confirmation for dangerous operations, or session/credential isolation is shown. README claims of local-only storage and content-free telemetry are self-reported and unverifiable from source; score 5.
Maintenance 10/20
The repo is not archived, has an AGPL-3.0 license, an identifiable team, and community support channels, showing ongoing maintenance. But the evidence gives no commit frequency, release/tag history, dependency-update record, or security-response channel, and the server package is still 0.0.1. Star count is only a discovery signal and was not scored. Active but with governance gaps: 10.
Documentation 8/20
The README offers substantial product-level content, downloads, MCP connection paths, FAQ, privacy notes, and links to official docs. But server-specific documentation lacks a tool list, parameter reference, limits/cost disclosure, troubleshooting, and verifiable evidence; users must work under hidden assumptions. Score 8.
Setup experience 7/20
Per the README, setup is download the BrowserOS neo desktop app, then connect via one-click or URL; server.json declares an npm/stdio package. But no concrete client config example, npm install command, or runtime prerequisite (e.g., app must be running first) is provided, and there is no execution evidence under static calibration, so the steps are not confirmed reproducible. Score 7.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessControls a browser

Best for

  • Developers who need agents to automate real workflows in logged-in sessions (e.g., personal email, social media, CRM)
  • Users who prefer local browser automation with data privacy
  • Those who need to run multiple AI agents in parallel and monitor their activity

Not for

  • CI/CD testing in stateless, login-free environments (use Playwright or browser-use instead)
  • Users who prefer cloud browser services over local execution
  • Those who don't need browser automation or only need simple web scraping

Required permissions

  • Read and import browser logins (from Chrome)
  • Access user browser sessions for agent automation
  • Store session recordings, screenshots, and history locally under ~/.browserclaw/
  • Send anonymous product usage events (agent connect/disconnect, version, OS) – never URLs, page content, prompts, tool results, or screenshots

Risks and side effects

  • Agents use your real login sessions and may perform unintended actions (e.g., sending emails, posting content) – approve with caution
  • Malicious prompts could lead to sensitive data exposure or unauthorized actions
  • Although data is stored locally, inherent risks of browser automation (e.g., mis-clicks, form errors) remain
  • Open-source version is AGPL-3.0 licensed, which may not suit all commercial use cases

Setup

Before you start

Runtime：Prebuilt binary

Install BrowserOS neo and import your logins from Chrome in one click, along with bookmarks and extensions. Then, in BrowserOS neo, connect your AI tools with one click (Claude Code, Codex, Cursor, VS Code, Zed, OpenCode, Antigravity) or via URL for any MCP client. For Claude Desktop, use the provided drop-in extension.

Check that it works

After installing BrowserOS neo, use the one-click MCP connect (or the URL endpoint) in an MCP client like Claude Code or Cursor and confirm the browser automation tools appear in the client's tool list, then give the agent a web task and watch it live in the neo dashboard.

Troubleshooting

  1. Ensure BrowserOS neo is installed and running
  2. Verify the agent is correctly connected to the MCP endpoint (often a local address)
  3. If agents can't access logins, confirm import from Chrome and granted permissions
  4. Check official docs (docs.browseros.com/neo) or seek help on Discord/Slack
  5. Update BrowserOS neo or MCP server to the latest version

Things to try

Once connected, you can ask your AI assistant things like:

  • Book me the cheapest flight to London.
  • Unsubscribe me from junk email newsletters.
  • Find apartments on a listings site and message the landlords.

Use cases

Let AI agents book flights and restaurants
Unsubscribe from junk email
Find apartments and message landlords
Reply on X, queue posts, pull engagement numbers
Update your CRM, file expenses, pull reports from internal tools

Supported clients

Claude Desktop
Claude Code
Codex
Cursor
VS Code
Zed
OpenCode
Antigravity

Listed from the project's documentation, not tested by this site.

Overview

BrowserOS MCP Server is a Model Context Protocol (MCP) server that exposes Chromium browser automation capabilities through the BrowserOS platform. It allows AI agents such as Claude Code, Codex, Cursor, and others to control the BrowserOS neo browser via the MCP protocol, leveraging real login sessions, parallel task execution, live monitoring, and session replay. Developed by the BrowserOS team, it is part of the BrowserOS open-source project.

Similar servers

Compare with other MCP servers

Source revision 53a85c9e2945 Data synced 2026-10-12 Read the FMRS scoring method