mcp-typescribe
23
MCP-Typescribe is an MCP Server providing Large Language Models (LLMs) with access to TypeScript API documentation in real-time. It enables LLMs to query, plan, and adapt to unfamiliar APIs without retraining, promoting faster onboarding and more autonomous AI-based coding assistants.
MCP-Typescribe
Overview
MCP-Typescribe is an MCP Server designed to provide Large Language Models (LLMs) with real-time access to API documentation, particularly focusing on TypeScript definitions. This project aims to overcome the limitations LLMs face when using unseen APIs and libraries by dynamically serving API context.
Features
- TypeDoc Integration: Efficiently loads and indexes TypeScript API JSON documentation.
- Comprehensive Query Tools: Offers a wide range of query capabilities for exploring TypeScript APIs.
- MCP Protocol: Seamlessly integrates with AI agents to provide real-time API context.
Query Capabilities
- Search for symbols and their details.
- List methods and properties of classes or interfaces.
- Find implementations and usages.
- Explore type hierarchies and search by return types.
Usage
To use MCP-Typescribe:
- Generate TypeDoc JSON for your TypeScript API.
- Build the project.
- Start the MCP server.
- Connect an AI agent to query the API.