Cocoindex

CocoIndex is an open source incremental data framework, with a Python API, for keeping the context that AI agents read continuously fresh. You declare the desired state of a target store, and the engine watches the sources, computes what changed and reconciles only that delta. When one file changes, one row is resynced; when the transformation code changes, only affected rows rerun and the schema evolves without an index swap or downtime. The vendor compares the model to React applied to data engineering. It is built for engineers building coding, code review, security review and other long running agents that break when their data goes stale. Sources can include codebases, meeting notes, pull request reviews and Slack. Working starter projects cover real time codebase indexing kept in sync with every commit, turning meeting notes into a live knowledge graph of people, topics, decisions and action items, and an LLM wiki that walks many git repositories and summarizes each one plus the organization as a whole. A companion project, CocoIndex Code, provides syntax tree based coding context: call graphs, hierarchies, symbol tables and semantic indexes that stay current as the repository changes, with semantic search by meaning and blast radius analysis before a change ships. No pricing page is published. The framework is open source and distributed through GitHub, and no paid tier is listed on the site.