Provides semantic code search by parsing source files with tree-sitter into functions, classes, and methods, then encoding them with vector embeddings stored in LanceDB. Supports local ONNX embeddings with ~100ms startup as well as cloud providers (OpenAI, Voyage AI, Google Gemini) via bring-your-own-key configuration. Returns the top semantically relevant code chunks instead of entire files, achieving significant token reduction for codebase queries. Supports incremental re-indexing, 10+ programming languages, and approximate nearest neighbor search via IVF-PQ indexing.
Cognium trust score
89%
Tier
Verified
Composite of vulnerability cleanliness, spec conformance, provenance, stability, and usage signals — scanned and weighted by Cognium. Human and agent signals are tracked separately.
Last scanned 2026-09-28.
Returns 7 tools: search_skills, get_skill, list_leaderboard, get_trust_breakdown, resolve_composition, plus the ChatGPT-connector search and fetch. Every tool is annotated read-only.
Resolve this skill directly via MCP tools/call get_skill.