enola-labs/enola

A lot of agent context tooling still relies on an LLM re-reading files and summarizing them, which drifts from ground truth as a codebase grows. Enola instead builds a deterministic graph straight from source across a dozen languages and frameworks, then lets agents query it like a database. That is a small but meaningful shift from probabilistic context to verifiable structure. Worth a look for anyone tired of watching agents burn tokens rediscovering the same import graph every session.