RTFM MCP Server

RTFM

Retrieve The Forgotten Memory

The open retrieval layer for AI agents

Index your entire project — code, docs, legal, research, data — and serve your AI agent exactly the context it needs.

Why?

Your AI agent is blind. It greps through thousands of files, loses context every session, hallucinates modules that don't exist. The fix isn't a smarter model — it's smarter retrieval.

Augment, Sourcegraph, and Cursor index code. RTFM indexes everything.

pip install rtfm-ai[mcp] && cd your-project && rtfm init

30 seconds. Claude Code now searches your indexed knowledge base before grepping.

Features

Search & Retrieval

• FTS5 full-text search — instant, zero-config, works out of the box
• Semantic search — optional embeddings (FastEmbed/ONNX, no GPU needed)
• Metadata-first — search returns file paths + scores (~300 tokens), not content dumps
• Progressive disclosure — the agent reads only what it needs via Read(file_path)

Indexing

• 10 parsers built-in — Markdown, Python (AST), LaTeX, YAML, JSON, Shell, PDF, XML, HTML, plain text
• Extensible — add any format in ~50 lines of Python
• Incremental sync — only re-indexes what changed
• Auto-sync — hooks keep the index fresh every prompt, zero manual work

Integration

• MCP server — works with Claude Code, Cursor, Codex, any MCP client
• CLI — rtfm search, rtfm sync, rtfm status, ...
• Python API — Library, SearchResults, custom parsers
• Non-invasive — doesn't touch your code, doesn't replace your workflow tools

Quick Start

Install

pip install rtfm-ai[mcp]

Initialize in your project

cd /path/to/your-project
rtfm init

This creates .rtfm/library.db, registers the MCP server, injects search instructions into CLAUDE.md, and installs auto-sync hooks. Done.

Then say to Claude Code: "Search for authentication flow" — it uses rtfm_search instead of grepping.

Optional extras

pip install rtfm-ai[embeddings]  # Semantic search (FastEmbed ONNX)
pip install rtfm-ai[pdf]         # PDF parsing (pdftext + marker)
pip install rtfm-ai[mcp,embeddings,pdf]  # Everything

MCP Tools

The Parser Architecture

This is what makes RTFM different. Need to index a format nobody supports?

from rtfm.parsers.base import BaseParser, ParserRegistry
from rtfm.core.models import Chunk

@ParserRegistry.register
class FHIRParser(BaseParser):
    """Parse HL7 FHIR medical records."""
    extensions = ['.fhir.json']
    name = "fhir"

    def parse(self, path, metadata=None):
        data = json.loads(path.read_text())
        for entry in data.get('entry', []):
            resource = entry.get('resource', {})
            yield Chunk(
                id=resource.get('id', str(uuid4())),
                content=json.dumps(resource, indent=2),
                book_title=f"FHIR {resource.get('resourceType', 'Unknown')}",
                book_slug=resource.get('id', 'unknown'),
                page_start=1,
                page_end=1,
            )

50 lines. Now your medical AI agent understands FHIR records.

Built-in parsers

| Parser | Extensions