browser-mcp
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@browser-mcpsearch for latest node.js version"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
@sthbryan/browser-mcp
Model Context Protocol server for real browser automation — Puppeteer + stealth, system browsers first.
Sibling to obscura-mcp-server: use Obscura for light scrape/text; use browser-mcp when you need pixels / JS render / stealth search.
Tools
screenshot
Capture a page with device templates or a custom size.
{
"url": "https://example.com",
"template": "mobile",
"fullPage": false,
"format": "png"
}Templates: mobile (390×844), tablet (768×1024), desktop (1440×900), custom (+ width/height).
fetch_page
Fetch a URL after JS render as html, markdown, or text.
{ "url": "https://example.com", "type": "markdown", "template": "desktop" }search
Web search with stealth. Engines: DuckDuckGo Lite → DDG HTML → Brave.
{ "query": "rust headless browser", "limit": 5 }query
Extract data with CSS selectors and/or text filters.
{ "url": "https://example.com", "selector": "h1" }{ "url": "https://example.com", "selector": "a", "attribute": "href", "limit": 10 }Browser resolution (lightweight)
No Chromium download on install. At runtime, in order:
BROWSER_MCP_EXECUTABLE_PATHif setSystem apps: Chrome → Chromium → Brave → Edge (override with
BROWSER_MCP_BROWSER)Local caches (Playwright/Puppeteer already on disk)
Last resort: download Chrome to
~/.cache/browser-mcp(disable withBROWSER_MCP_ALLOW_DOWNLOAD=0)
Always headless + puppeteer-extra-plugin-stealth.
Install (MCP client)
{
"mcpServers": {
"browser-mcp": {
"command": "bunx",
"args": ["-y", "@sthbryan/browser-mcp"]
}
}
}Tool filter:
bunx @sthbryan/browser-mcp --tools=screenshot
bunx @sthbryan/browser-mcp --tools=search,querySee INSTALL.md.
Requirements
Node 18+ or Bun
A Chromium-based browser installed (recommended): Chrome / Brave / Edge
export BROWSER_MCP_BROWSER=brave
export BROWSER_MCP_EXECUTABLE_PATH=/path/to/chrome
export BROWSER_MCP_ALLOW_DOWNLOAD=0
export BROWSER_MCP_ALLOW_PRIVATE=1Release / publish
CI (same pattern as obscura): on every PR/main push → lint, test, smoke, build.
On tag v* → publish to npm + GitHub Release.
# 1) Ensure GitHub secret NPM_TOKEN (classic publish token for @sthbryan)
# 2) From a clean main branch:
bun run release # tags current package.json (0.1.0) and pushes
bun run release patch # bump + tag + push
bun run release --no-pushTag must match package.json version (e.g. v0.1.0 ↔ 0.1.0).
Architecture
MCP Client ──stdio──► browser-mcp
│
▼
resolve system browser
(Chrome/Brave/Edge/…)
│
▼
puppeteer-core + stealth
│
┌───────────┴───────────┐
▼ ▼
screenshot search
fetch_page (DDG→Brave)
queryDevelopment
bun install
bun test
bun run smoke
bun run check
bun run buildLicense
MIT
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/sthbryan/browser-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server