Lancer

关于

Lancer MCP is a self-hosted Model Context Protocol (MCP) server designed to empower AI agents with advanced code intelligence. It efficiently clones and tracks Git repositories, performing incremental indexing, multi-language code parsing, and embedding generation. By combining BM25 full-text search, vector semantic search, and graph re-ranking, Lancer delivers highly relevant search results. Its unique approach utilizes a single, unified `Query` MCP tool to optimize performance and accuracy for LLMs, offering a private, powerful, and extensible alternative to cloud-based code indexing services.

主要功能

• Hybrid code search combining BM25, vector embeddings, and graph traversal.
• Unified `Query` MCP tool for efficient AI agent interaction and optimized function calling.
• 0 GitHub stars
• Self-hosted MCP server for privacy and control over code intelligence.
• Multi-language code parsing with Roslyn for C# and regex for other languages.
• Git repository tracking with incremental indexing and persistent state.

使用案例

• Powering personal LLMs with high-quality, contextually relevant code search without cloud dependencies.
• Creating a LAN-accessible, private platform for code indexing and repository intelligence.
• Improving LLM reasoning and accuracy by providing a single, optimized function calling interface for code exploration.