#code-search

MCP-native code retrieval for AI agents — 84-88% fewer read tokens, BM25F + semantic search, AST chunks, session dedup

Local-first AI agent for code, audit, review, web search and research — in Go. Own agent loop, memory, skills, codebase index and TUI. Ollama/llama.cpp by default; opt-in cloud APIs.

Sourcegraph is a code intelligence platform that lets developers and AI agents search, navigate, understand, and modify code across very large, multi-repo codebases. The platform pairs classic code search and code intelligence (jump-to-definition, references, batch changes) with AI products — Cody (in-IDE AI coding assistant), Amp (frontier coding…

Indexed code navigation for AI coding agents (Claude Code, Cursor, any MCP client) — replaces grep scans with GNU Global (gtags) lookups. ~100x faster, radically less context noise. One-line no-sudo install, everything in user space.

Apex-relevance subgraph retrieval for AI agents — feed your LLM the most relevant slice of a knowledge graph, within a token budget.

MCP server giving coding agents semantic + structural search over your codebase, so they reuse proven patterns instead of guessing. Finalist - Cursor × Superteam Georgia Hackathon 2026.