Show HN: Ragpilot – MCP that cuts AI agent context tokens 6-9x via local RAG

Category: ai-ml

Tags: rag, mcp-server, code-search, ai-tools, rust

Score: 7.3/10 (Innovation: 7, Technical: 8, Documentation: 7, Utility: 7)

Ragpilot is a Rust-based MCP server that provides RAG-powered tools for AI agents to understand local codebases, featuring semantic search, symbol graph, impact analysis, and context bundling with token budgeting. Its innovative combination of local vector search (Qdrant), multi-language parsing via Tree-sitter, and per-client MCP registration significantly reduces context token usage for AI coding assistants. The project stands out for its practical approach to optimizing LLM context windows and its reproducible benchmark demonstrating 6-9x token savings.

Target audience: backend devs, data engineers, ai engineers

Repository: https://github.com/alikaya/ragpilot · Rust · MIT

View on Hacker News