cocoindex

Why we included this project

Keeping a RAG or agent-memory index current usually means re-embedding everything on every change, which gets expensive as your corpus grows. CocoIndex takes a different path: you declare the target state as a function of your source data, and its Rust engine watches the sources and reprocesses only the delta. Codebases, Slack conversations, docs, and meeting notes stay continuously in sync with the agents that read them, without full re-indexing cycles, and the engine keeps lineage so you can trace where a piece of context came from. You write pipelines in plain declarative Python rather than wiring orchestration DAGs, which makes it usable for data engineers who are not already pipeline specialists. That combination is what you want when you run production LLM apps, semantic search, or long-horizon agents and your retrieval layer has to reflect the latest state.