Browsegrab
Playwright meets accessibility trees for browser automation through Claude. You get 8 tools: navigate, click, type, scroll, snapshot, extract_content, go_back, and wait. The hook here is token efficiency for local LLMs. Instead of dumping full HTML to your context window, it sends accessibility trees with ref IDs like e1, e2 for stable element targeting, plus MarkGrab for clean markdown extraction. Claims 500 to 1,500 tokens per step versus 4,000 to 10,000 for browser-use. Built by the QuartzUnit folks who made MarkGrab, SnapGrab, and DocPick. Reach for this when you need Claude to navigate websites but don't want to burn tokens on raw DOM dumps or coordinate hunting.
- tools
- 8
- transport
- STDIO
- last updated
- Mar 18, 2026
Tools
8 toolsPublic tool metadata for what this MCP can expose to an agent.
browser_navigate
No parameter schema in public metadata yet.
browser_click
No parameter schema in public metadata yet.
browser_type
No parameter schema in public metadata yet.
browser_snapshot
No parameter schema in public metadata yet.
browser_scroll
No parameter schema in public metadata yet.
browser_extract_content
No parameter schema in public metadata yet.
browser_go_back
No parameter schema in public metadata yet.
browser_wait
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageReadme
browsegrab
Token-efficient browser agent for local LLMs — Playwright + accessibility tree + MarkGrab, MCP native.
browsegrab is a lightweight browser automation library designed for local LLMs (8B-35B parameters). It combines Playwright's accessibility tree with MarkGrab's HTML-to-markdown conversion to achieve 5-8x fewer tokens per step compared to alternatives like browser-use.
Features
- Token-efficient: ~500-1,500 tokens/step (vs 4,000-10,000 for browser-use)
- Local LLM first: Optimized for vLLM, Ollama, and OpenAI-compatible endpoints
- MCP native: Built-in MCP server with 8 browser automation tools
- MarkGrab integration: HTML → clean markdown for content extraction
- Accessibility tree + ref system: Stable element references (
e1,e2, ...) without vision models - Success pattern caching: Zero LLM calls on repeated workflows
- 5-stage JSON parser: Robust action parsing for local LLM outputs
- Minimal dependencies: Only
playwright+httpxin core
Installation
pip install browsegrab
playwright install chromium
Details
- Repository
- github.com/quartzunit/browsegrab
- Categories
- Web & Browser Automation, AI & LLM Tools
- Package
browsegrab- Transport
- STDIO
- Registry
- MCP registry, active
More from quartzunit
Related Web & Browser Automation MCP servers
View all
BrowserToolsagentdeskai/browser-tools-mcpMonitor browser logs directly from Cursor and other MCP compatible IDEs.7.2Kstars
Openbrowser Aibilly-enrizky/openbrowser-aiAI browser automation. Write async Python to navigate, click, type, and extract data.279stars
One Search Mcpyokingma/one-search-mcpWeb search, URL discovery, scraping & extraction with agent-browser and search providers148stars
Native DevToolssh3ll3x3c/native-devtools-mcpMCP server for desktop, browser (CDP), and Android automation. Screenshot, OCR, click, type.113stars
Mcp Accessibility Scannerjustasmonkev/mcp-accessibility-scannerMCP server for automated web accessibility scanning with Playwright and Axe-core.54stars
Openevidence Mcpbakhtiersizhaev/openevidence-mcpUnofficial MCP server connecting OpenEvidence to AI agents via authenticated browser sessions49stars