Chrome Browser
BLOCKgrade F · trust 43/100Chrome MCP Server is a Chrome extension-based Model Context Protocol (MCP) server that exposes your Chrome browser functionality to AI assistants like Claude, enabling complex browser automation, content analysis, and semantic search.
Overview
From the repository's own README, as read at the audited commit.
# Chrome MCP Server 🚀[](https://img.shields.io/github/stars/hangwin/mcp-chrome)[](https://opensource.org/licenses/MIT)[](https://www.typescriptlang.org/)[](https://developer.chrome.com/docs/extensions/)[](https://img.shields.io/github/v/release/hangwin/mcp-chrome.svg)> 🌟 **Turn your Chrome browser into your intelligent assistant** - Let AI take control of your browser, transforming it into a powerful AI-controlled automation tool.**📖 Documentation**: [English](README.md) | [中文](README_zh.md)> The project is still in its early stages and is under intensive development. More features, stability improvements, and other enhancements will follow.---## 🎯 What is Chrome MCP Server?Chrome MCP Server is a Chrome extension-based **Model Context Protocol (MCP) server** that exposes your Chrome browser functionality to AI assistants like Claude, enabling complex browser automation, content analysis, and semantic search. Unlike traditional browser automation tools (like Playwright), **Chrome MCP Server** directly uses your daily Chrome browser, leveraging existing user habits, configurations, and login states, allowing various large models or chatbots to take control of your browser and truly become your everyday assistant.## ✨ New Features(2025/12/30)- **A New Visual Editor for Claude Code & Codex**, for more detail here: [VisualEditor](docs/VisualEditor.md)## ✨ Core Features- 😁 **Chatbot/Model Agnostic**: Let any LLM or chatbot client or agent you prefer automate your browser- ⭐️ **Use Your Original Browser**: Seamlessly integrate with your existing browser environment (your configurations, login states, etc.)- 💻 **Fully Local**: Pur
Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add wasm-simd --env ANTHROPIC_AUTH_TOKEN=${ANTHROPIC_AUTH_TOKEN} --env CHROME_EXTENSION_KEY=${CHROME_EXTENSION_KEY} -- npx -y @chrome-mcp/[email protected]{
"mcpServers": {
"wasm-simd": {
"command": "npx",
"args": [
"-y",
"@chrome-mcp/[email protected]"
],
"env": {
"ANTHROPIC_AUTH_TOKEN": "${ANTHROPIC_AUTH_TOKEN}",
"CHROME_EXTENSION_KEY": "${CHROME_EXTENSION_KEY}"
}
}
}
}Exposed tools (7) 7 read · 0 write · 0 destructive
| Tool | Risk | Description |
|---|---|---|
Auto | read | Cursor auto-selects the best model |
GPT-4.1 | read | OpenAI model via Cursor |
GPT-5.1 | read | OpenAI high-quality reasoning model |
GPT-5.2 | read | OpenAI flagship reasoning model with extended effort support |
Test | read | A test flow |
__MSG_extensionName__ | read | __MSG_extensionDescription__ |
url | read | Target URL |
Details
- Source
- hangwin/mcp-chrome
- npm
@chrome-mcp/wasm-simd@0.1.0- Transports
- sse · stdio · streamable-http
- Credentials it reads
ANTHROPIC_AUTH_TOKENCHROME_EXTENSION_KEY- License
- MIT
- Stars
- 12,405 · pushed 249d ago
Trust audit
Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | FAIL |
| L1 | Static analysis of the code | FAIL |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (10 observation(s))
- Network
- declared (10 observation(s))
- Shell
- declared (10 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (25)
payload-builder.ts
const fn = new Function(code);
const fn = new Function(...argNames, wrappedCode);
const fn = new Function(
const fn = new Function(
func: (code) => new Function(code)(),
{ color = CONFIG.COLORS.HOVER, dashed = true, offsetX = 0, offsetY = 0, isVerify = false } = {},function drawRects(elements, color = CONFIG.COLORS.HOVER, dashed = true, isVerify = false) {simd_math_bg.wasm
const url = `http://127.0.0.1:${port}/agent/sessions/${encodeURIComponent(sessionId)}`;const sseUrl = `http://127.0.0.1:${request.port}/agent/chat/${encodeURIComponent(request.sessionId)}/stream`;const url = `http://127.0.0.1:${request.port}/agent/chat/${encodeURIComponent(request.sessionId)}/act`;.prettierignore
.prettierrc.json
import type { NodeId, RunId } from '../../domain/ids';import type { RRError } from '../../domain/errors';import { RR_ERROR_CODES, createRRError } from '../../domain/errors';import type { NodeId, RunId } from '../../domain/ids';import type { Breakpoint, DebuggerState } from '../../domain/debug';"url": "http://127.0.0.1:12306/mcp"
"url": "http://127.0.0.1:12306/mcp"
const byteChars = atob(base64Data);
const decodedString = atob(base64);
@modelcontextprotocol/sdk, @vue-flow/background, @vue-flow/controls, @vue-flow/core, @vue-flow/minimap, @xenova/transformers, date-fns, elkjs
@anthropic-ai/claude-agent-sdk, @fastify/cors, @modelcontextprotocol/sdk, better-sqlite3, chalk, chrome-devtools-frontend, commander, drizzle-orm
Gates applied: no_behavioural_pass.
Audited 2026-09-13 · audit v0.4.0 · source sha 54eb663ae58f · full audit: observations/trust-audit/mcp-server/hangwin__chrome-browser.json · Report an issue or request a re-scan
Audit history
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-09-13 | 54eb663ae58f | BLOCK | F | 43 | first audit |
Also audited
Servers with a similar trust profile.
Questions
What is the Chrome Browser MCP server?
Chrome MCP Server is a Chrome extension-based Model Context Protocol (MCP) server that exposes your Chrome browser functionality to AI assistants like Claude, enabling complex browser automation, content analysis, and semantic search.
What tools does Chrome Browser expose?
7 in total: 7 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is Chrome Browser safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (43/100) and found 8 critical or high issues in the source. Each one is listed on this page with the file and line it is on.
What credentials does Chrome Browser need?
It reads ANTHROPIC_AUTH_TOKEN and CHROME_EXTENSION_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does Chrome Browser run?
It speaks sse, stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as @chrome-mcp/wasm-simd at 0.1.0.
How current is this page?
The grade is for one exact copy of the source (54eb663ae58f), read on 2026-09-13. The repository is watched and re-audited when it changes.
Provenance: OBSERVED · read 2026-09-13 · job trust-audit-2026-09-13