doc-fetcher

name	doc-fetcher
description	Fetch external documentation sites and convert to local markdown for context
user_invocable	true

Doc Fetcher

Fetch and convert external documentation to local markdown files for use as context (RAG, reference, etc.).

Prerequisites

CF Browser MCP server must be configured in .mcp.json (see quick start).

Workflow

Step 1: Discover Pages

browser_links(base_url) → list of all linked pages

• Filter to same-domain URLs only
• Filter by documentation patterns (e.g., /docs/, /guide/, /api/, /reference/)
• Cap at max_pages (default 20)

Step 2: Fetch Content

For each discovered page:

browser_markdown(page_url) → markdown content

Process pages in parallel where possible (batch of 5).

Step 3: Clean Content

For each page:

• Strip navigation headers/sidebars
• Remove footer boilerplate
• Remove cookie banners and ads
• Preserve code blocks and tables

Step 4: Save to Disk

Save to docs/{source_domain}/:

docs/
└── docs.astro.build/
    ├── index.md          ← Table of contents
    ├── getting-started.md
    ├── routing.md
    └── components.md

Step 5: Generate Index

Create index.md with:

• Source URL and fetch timestamp
• Table of contents with links to each file
• Total page count and approximate token count

Step 6: Report

Output summary:

• Pages fetched successfully
• Total approximate token count
• Key topics covered
• Any failed pages

Parameters

• url (required): Base URL of documentation site
• max_pages (optional, default=20): Maximum pages to fetch
• output_dir (optional, default="docs/{domain}"): Where to save files
• patterns (optional): URL patterns to include (e.g., "/docs/", "/api/")

Examples

"Fetch the Hono documentation"
→ browser_links("https://hono.dev") → filter /docs/ → browser_markdown each → save

"Download Cloudflare Workers docs for reference"
→ browser_links("https://developers.cloudflare.com/workers/") → fetch up to 20 pages

"Index the Astro guide for RAG"
→ browser_links("https://docs.astro.build/") → fetch + clean + save with index