ontomics

ontomics gives a coding agent instant knowledge of a codebase's domain concepts through a single MCP tool call, rather than the dozens of searches an agent would otherwise need. In a benchmark on an open-source image-registration library, answering what a key term means in that codebase took one call and about 3,700 tokens with ontomics against nineteen calls and roughly 76,000 tokens without it, with equally complete answers in both cases.

It builds a semantic index by clustering related symbols into concepts, detecting naming conventions from usage frequency, resolving abbreviations, and embedding raw function bodies to cluster them by behavioral similarity, which can surface duplicated implementations that neither naming nor call graphs reveal. That index can be exported as a portable artifact to bootstrap conventions in another repository. It requires a git repository, auto-detects and indexes it on first run, and is written in Rust.