@recon-fuzz-mcp/knowledge

MCP server that makes Recon documentation searchable by AI tools. Fetches and parses getrecon.xyz/llms-full.txt into structured, queryable content.

Tools

Site tools (getrecon.xyz)

Book tools (book.getrecon.xyz)

Substack tools (getrecon.substack.com)

Cross-source

Installation

Claude Code

claude mcp add recon-knowledge -- npx @recon-fuzz-mcp/knowledge

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "recon-knowledge": {
      "command": "npx",
      "args": ["@recon-fuzz-mcp/knowledge"]
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "recon-knowledge": {
      "command": "npx",
      "args": ["@recon-fuzz-mcp/knowledge"]
    }
  }
}

No API key needed. The server fetches public documentation only.

Local development

git clone https://github.com/Recon-Fuzz/recon-mcp-knowledge.git
cd recon-mcp-knowledge
npm install
npm run build

Test it works

# List tools
echo '{"jsonrpc":"2.0","method":"tools/list","id":1}' | node dist/index.js

# Search for chimera content
echo '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"search_site","arguments":{"query":"chimera"}},"id":2}' | node dist/index.js

Architecture

• Fetches llms-full.txt once on startup, caches in memory

• Cache refreshes every 24h automatically or on manual refresh_cache

• Parser splits by --- dividers, extracts blog posts, glossary, comparisons, tools

• Search uses case-insensitive term matching with word-boundary scoring

• No database, no external search library, no filesystem writes

Privacy

This server is read-only. It fetches from a single hardcoded public URL (getrecon.xyz/llms-full.txt). No user queries, tool arguments, or any data is sent to Recon or any third party.