LLM-Powered Scraping
7-step pipeline with LLM-assisted scope analysis, CSS selector detection, and automatic SPA support via Selenium.
Scrape any documentation site and inject it into your AI coding agent's context.
Works with Windsurf, Cursor, Claude Code, and Antigravity.
Everything you need to supercharge your AI assistant with any documentation
7-step pipeline with LLM-assisted scope analysis, CSS selector detection, and automatic SPA support via Selenium.
Native integration with Windsurf, Cursor, Claude Code, and Antigravity. Each IDE gets docs injected in its own format.
~500 byte always-on index per doc set. The AI agent sees the full file tree and reads individual docs on-demand.
Docs stored globally in AppData, shared across projects. No cloud dependency. Your docs stay on your machine.
Automatic detection of client-side rendered sites. Seamless Selenium fallback for React, Angular, and Next.js docs.
Automatic removal of repeated UI artifacts across pages. Clean markdown output ready for AI consumption.
Ask your AI agent to scrape a docs site. It returns a CLI command that runs the 7-step pipeline.
scrape_docs(
url="https://docs.example.com",
name="example-docs"
)
Add the docs to your project. Files are copied and a compact index is injected as an always-on IDE rule.
add_docs_to_project(
name="example-docs",
project_root="/my/project"
)
The AI agent now sees the doc index in every conversation and reads individual files on-demand. No extra prompting needed.
// Agent sees ~500 byte index
// Reads individual files as needed
"How do I configure auth?"
Native integration with 4 AI coding agents
.windsurf/rules/
.cursor/rules/
CLAUDE.md
.agent/rules/
Inspired by @gaojude/next-agents-md and Vercel's research on retrieval-led reasoning.
A compressed docs index in AGENTS.md achieved 100% pass rate vs 79% for skills. AnyDocsMCP applies this pattern to any documentation site.
Open source and free forever. Start using AnyDocsMCP in under 5 minutes.