Llmmcp

io.github.zero-abdPublished by GitHub account zero-abdv1.2.0Updated Oct 24, 2026

MCP server for real-time LLM API documentation — stop hallucinations

VerifiedSTDIODesktop onlyKnowledge & MemoryDeveloper ToolsAI & ML

Overview

AI-generated overview

Local MCP server that gives an assistant real-time, up-to-date documentation for OpenAI, Anthropic and Google Gemini APIs via a hosted search backend.

What you can do
  • Search official LLM provider documentation for technical details such as tool-use parameters, context windows and rate limits
  • List available providers and their currently promoted models from a dynamically updated source
  • Ground answers about model versions, feature availability and pricing in indexed official docs instead of training data
When to use it
Useful when an assistant writes code against OpenAI, Anthropic or Gemini APIs and you want it to check current model names, parameters and patterns rather than rely on possibly outdated training data.
Requirements
Runs as a local stdio process, started with npx (Node.js required). No API keys or environment variables are declared; the CLI calls a hosted Cloudflare Worker backend, so network access is needed.
Before you install
Queries are sent to a third-party hosted backend (Cloudflare Worker with a Pinecone vector index), so search terms leave your machine. Documentation is refreshed by a weekly scraper, so it may lag behind the very latest provider changes.

Installation

In SourceWeft

  1. Open Llmmcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

llmmcp

[MCP Registry]

🌐 Website: https://llmmcp.vercel.app
🎥 Demo:

https://github.com/user-attachments/assets/eaad8d05-b7a8-4bf0-86c6-4fe2726da628

Stop LLM hallucinations and outdated code patterns.

llmmcp is a Model Context Protocol (MCP) server that provides real-time, up-to-date documentation for major LLM providers (OpenAI, Anthropic, and Google Gemini). It ensures your AI agents—like Cursor, Claude Desktop, or Windsurf—base their work on current official documentation instead of stale training data or deprecated library patterns.

Why use llmmcp?

LLMs frequently hallucinate about their own latest versions, feature availability (e.g., tool use in certain models), and pricing. llmmcp fixes this by providing:

  • ✅ Up-to-Date Model Info: Always know the latest available models (e.g., Gemini 2.0 Flash, Claude 3.5 Sonnet).
  • ✅ Detailed API Params: Verified tool use syntax, context window sizes, and rate limits.
  • ✅ Latest Implementation Patterns: Force your AI agent to follow current best practices instead of using legacy or deprecated library versions.
  • ✅ Real-Time Search: Queries an indexed vector database of official provider documentation.
  • ✅ Dynamic Listings: Get the current state of providers without hardcoded lists.

🚀 Quick Start

You can use llmmcp immediately in your favorite AI tools without local installation.

Cursor

Add a new MCP server in Settings > Models > MCP Servers:

  • Name: llmmcp
  • Type: command
  • Command: npx -y llmmcp@latest

Claude Desktop

Add the following to your claude_desktop_config.json:

json
{  "mcpServers": {    "llmmcp": {      "command": "npx",      "args": ["-y", "llmmcp@latest"]    }  }}

🛠 Features

search_docs

Search the latest official documentation for specific technical details. Example: "What are the tool use parameters for Gemini 1.5 Pro?"

list_providers

Get a dynamically updated list of available providers (OpenAI, Anthropic, Google) and their currently promoted models.

🏗 How it Works

llmmcp is designed for speed and reliability:

  1. Indexer: A weekly scraper fetches raw markdown/text from official documentation.
  2. Vector DB: Chunks are embedded and stored in Pinecone with integrated embedding support.
  3. Backend: A Cloudflare Worker handles query embedding and retrieval, caching frequent results in Workers KV.
  4. MCP Client: A thin CLI translates MCP requests into API calls for the Worker.

🤝 Contributing & Self-Hosting

This project is open-source. If you'd like to run your own instance of the backend:

  1. See Architecture & Deployment (coming soon, see current setup in logs).
  2. Fork the repo and submit a PR for new documentation sources.

Developed by Abdullah Al Mahmud

License

MIT

Source: README.md at commit fdbd216

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.2.0LatestSep 16, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.