rtfd-mcp 0.2.1

MCP server for real-time library documentation access across multiple package ecosystems.

RTFD (Read The F*****g Docs) MCP Server

The RTFD (Read The F*****g Docs) MCP Server acts as a bridge between Large Language Models (LLMs) and real-time documentation. It allows coding agents to query package repositories like PyPI, npm, crates.io, GoDocs, DockerHub, GitHub, and Google Cloud Platform (GCP) to retrieve the most up-to-date documentation and context.

This server solves a common problem where LLMs hallucinate APIs or provide outdated code examples because their training data is months or years old. By giving agents access to the actual documentation, RTFD ensures that generated code is accurate and follows current best practices.

Why use RTFD?

• Accuracy: Agents can access the latest documentation for libraries, ensuring they use the correct version-specific APIs and avoid deprecated methods.
• Context Awareness: Instead of just getting a raw text dump, the server extracts key sections like installation instructions, quickstart guides, and API references, giving the agent exactly what it needs.
• Privacy: Unlike cloud-based documentation services, RTFD runs entirely on your local machine. Your queries are sent DIRECTLY to the source (no servers in the middle, no API keys needed, etc) and the documentation you access never leave your system, ensuring complete privacy and no data collection.
• Supported Sources: PyPI (Python), npm (JavaScript/TypeScript), crates.io (Rust), GoDocs (Go), Zig docs, DockerHub, GitHub repositories, and Google Cloud Platform (GCP).

Use Cases

RTFD helps in scenarios like:

• Refactoring old code: Fetch current pandas docs to find deprecated methods and their replacements. Instead of guessing what changed, the LLM reads the actual upgrade guide.

• Unfamiliar libraries: Integrating a Rust crate you've never seen? Look up the exact version, feature flags, and examples directly from the docs instead of guessing the API from general patterns.

• Libraries after training cutoff: Using a library released after the LLM's training data ends? Fetch the actual README and code examples from GitHub so the LLM can write correct usage instead of hallucinating APIs.

• Docker optimization: When optimizing a Dockerfile, inspect the official python:3.11-slim image to see exactly what packages and OS layers are included, rather than making assumptions.

• Dependency audits: Check PyPI, npm, and crates.io for available updates across all your dependencies. The LLM sees the latest versions and can generate an audit report without manually visiting each registry.

Features

• Documentation Content Fetching: Retrieve actual documentation content (README and key sections) from PyPI, npm, and GitHub rather than just URLs.
• Smart Section Extraction: Automatically prioritizes and extracts relevant sections such as "Installation", "Usage", and "API Reference" to reduce noise.
• Format Conversion: Automatically converts reStructuredText and HTML to Markdown for consistent formatting and easier consumption by LLMs.
• Multi-Source Search: Aggregates results from PyPI, npm, crates.io, GoDocs, Zig docs, DockerHub, GitHub, and GCP.
• Pluggable Architecture: Easily add new documentation providers by creating a single provider module.
• Error Resilience: Failures in one provider do not crash the server; the system is designed to degrade gracefully.

Installation

From PyPI (Recommended)

pip install rtfd-mcp

Or with uv:

uv pip install rtfd-mcp

From source

Clone the repository and install:

git clone https://github.com/aserper/RTFD.git
cd RTFD
pip install .
# or: uv pip install -e .

Quickstart

1. Export a GitHub token to avoid strict rate limits (optional but recommended):

   export GITHUB_TOKEN=ghp_your_token_here
   

2. Run the server:

   rtfd
   

3. Configure Documentation Fetching (Optional): Content fetching tools are enabled by default. To disable them and only use metadata tools:

   export RTFD_FETCH=false
   rtfd
   

4. Configure Token Counting (Optional): To enable token counting in response metadata (useful for debugging usage):

   export RTFD_TRACK_TOKENS=true
   rtfd
   

5. Configure Caching (Optional): Caching is enabled by default to improve performance and reduce load on providers.

   • Enable/Disable: export RTFD_CACHE_ENABLED=false (default: true)
   • TTL: export RTFD_CACHE_TTL=3600 (default: 604800 seconds / 1 week)
   • Location: ~/.cache/rtfd/cache.db

Releases & Versioning

For maintainers, see CONTRIBUTING.md for the automated release process.

To release a new version:

1. Go to Actions tab on GitHub
2. Select "Release to PyPI" workflow
3. Click "Run workflow"
4. Select version bump type (patch, minor, or major)
5. Done! The workflow will:
   • Update version in code
   • Create a git tag and commit
   • Create a GitHub release
   • Automatically publish to PyPI

For local version management:

python scripts/bump_version.py patch   # 0.1.0 → 0.1.1
python scripts/bump_version.py minor   # 0.1.0 → 0.2.0
python scripts/bump_version.py major   # 0.1.0 → 1.0.0

Available Tools

All tool responses are returned in JSON format.

Aggregator

• search_library_docs(library, limit=5): Combined lookup across all providers (PyPI, npm, crates.io, GoDocs, GCP, GitHub). Note: Zig and DockerHub are accessed via dedicated tools.

Cache Management

• get_cache_info(): Get cache statistics including entry count, database size, and location.
• get_cache_entries(): Get detailed information about all cached items including age, size, and content preview.

Documentation Content Fetching

• fetch_pypi_docs(package, max_bytes=20480): Fetch Python package documentation from PyPI.
• fetch_npm_docs(package, max_bytes=20480): Fetch npm package documentation.
• fetch_godocs_docs(package, max_bytes=20480): Fetch Go package documentation from godocs.io (e.g., 'github.com/gorilla/mux').
• fetch_gcp_service_docs(service, max_bytes=20480): Fetch Google Cloud Platform service documentation from docs.cloud.google.com (e.g., "storage", "compute", "bigquery").
• fetch_github_readme(repo, max_bytes=20480): Fetch README from a GitHub repository (format: "owner/repo").
• fetch_docker_image_docs(image, max_bytes=20480): Fetch Docker image documentation and description from DockerHub (e.g., "nginx", "postgres", "user/image").
• fetch_dockerfile(image): Fetch the Dockerfile for a Docker image by parsing its description for GitHub links (best-effort).

Metadata Providers

• pypi_metadata(package): Fetch Python package metadata.
• npm_metadata(package): Fetch JavaScript package metadata.
• crates_metadata(crate): Get Rust crate metadata.
• search_crates(query, limit=5): Search Rust crates.
• godocs_metadata(package): Retrieve Go package documentation.
• search_gcp_services(query, limit=5): Search Google Cloud Platform services by name or keyword (e.g., "storage", "compute", "bigquery").
• zig_docs(query): Search Zig documentation.
• docker_image_metadata(image): Get DockerHub Docker image metadata (stars, pulls, description, etc.).
• search_docker_images(query, limit=5): Search for Docker images on DockerHub.
• github_repo_search(query, limit=5, language="Python"): Search GitHub repositories.
• github_code_search(query, repo=None, limit=5): Search code on GitHub.

Integration with Claude Code

Add the following to your ~/.claude/settings.json:

{
  "mcpServers": {
    "rtfd": {
      "command": "rtfd",
      "type": "stdio"
    }
  }
}

Or with environment variables:

{
  "mcpServers": {
    "rtfd": {
      "command": "bash",
      "args": ["-c", "export GITHUB_TOKEN=your_token_here && rtfd"],
      "type": "stdio"
    }
  }
}

Pluggable Architecture

The RTFD server uses a modular architecture. Providers are located in src/RTFD/providers/ and implement the BaseProvider interface. New providers are automatically discovered and registered upon server restart.

To add a custom provider, create a new file in the providers directory inheriting from BaseProvider, implement the required methods, and the server will pick it up automatically.

Provider-Specific Notes

GCP (Google Cloud Platform)

• Service Discovery: Uses both a local service mapping (20+ common services) and GitHub API search of the googleapis/googleapis repository.
• Documentation Source: Fetches documentation by scraping docs.cloud.google.com and converting to Markdown.
• GitHub Token: Optional but recommended. Without a GITHUB_TOKEN, GitHub API search is limited to 60 requests/hour. With a token, the limit increases to 5,000 requests/hour.
• Supported Services: Cloud Storage, Compute Engine, BigQuery, Cloud Functions, Cloud Run, Pub/Sub, Firestore, GKE, App Engine, Cloud Vision, Cloud Speech, IAM, Secret Manager, and more.
• Service Name Formats: Accepts various formats (e.g., "storage", "cloud storage", "Cloud Storage", "kubernetes", "k8s" for GKE).

Other Providers

• Token Counting: Disabled by default. Set RTFD_TRACK_TOKENS=true to see token stats in Claude Code logs.
• Rate Limiting: The crates.io provider respects the 1 request/second limit.
• Dependencies: mcp, httpx, beautifulsoup4, markdownify, docutils, tiktoken.

Architecture

• Entry point: src/RTFD/server.py contains the main search orchestration tool. Provider-specific tools are in src/RTFD/providers/.
• Framework: Uses mcp.server.fastmcp.FastMCP to declare tools and run the server over stdio.
• HTTP layer: httpx.AsyncClient with a shared _http_client() factory that applies timeouts, redirects, and user-agent headers.
• Data model: Responses are plain dicts for easy serialization over MCP.
• Serialization: Tool responses use serialize_response_with_meta() from utils.py.
• Token counting: Optional token statistics in the meta field (disabled by default). Enable with RTFD_TRACK_TOKENS=true.

Serialization and Token Counting

Tool responses are handled by serialize_response_with_meta() in utils.py:

• Token statistics: When RTFD_TRACK_TOKENS=true, the response includes a _meta field with token counts (tokens_json, tokens_sent, bytes_json).
• Token counting: Uses tiktoken library with cl100k_base encoding (compatible with Claude models).
• Zero-cost metadata: Token statistics appear in the _meta field of CallToolResult, which is visible in Claude Code's special metadata logs but NOT sent to the LLM, costing 0 tokens.

Extensibility & Development

Adding Providers

The RTFD server uses a modular architecture. Providers are located in src/RTFD/providers/ and implement the BaseProvider interface. New providers are automatically discovered and registered upon server restart.

To add a custom provider:

1. Create a new file in src/RTFD/providers/.
2. Define async functions decorated with @mcp.tool().
3. Ensure tools return CallToolResult using serialize_response_with_meta(result_data).

Development Notes

• Dependencies: Declared in pyproject.toml (Python 3.10+).
• Testing: Use pytest to run the test suite.
• Environment: If you change environment-sensitive settings (e.g., GITHUB_TOKEN), restart the rtfd process.