GitHub - zilliztech/claude-context: Code search MCP for Claude Code. Make entire codebase the context for any coding agent.
5 hours ago
- #AI Coding Assistant
- #Semantic Code Search
- #MCP Plugin
- Claude Context is an MCP plugin that provides semantic code search for AI coding agents, giving them deep context from entire codebases.
- It uses a vector database (like Zilliz Cloud) and OpenAI embeddings for cost-effective, large-scale code indexing and retrieval.
- The plugin supports various AI assistants and IDEs through MCP, including Claude Code, VS Code, Cursor, and more, with detailed configuration examples provided.
- Key features include hybrid code search (BM25 + dense vector), incremental indexing, AST-based code chunking, and scalability via Zilliz Cloud.
- The core package can be used directly for indexing and searching, and a VS Code extension is available for semantic code search in the IDE.
- System requirements include Node.js 20.x or 22.x, and the project supports multiple languages, embedding providers, and vector databases.
- The evaluation shows ~40% token reduction with equivalent retrieval quality, leading to cost and time savings in production.