honcho

Why we included this project

Teams building assistants or agents that need to remember what happened in earlier conversations will find Honcho saves them a lot of plumbing. It stores messages and events, runs background reasoning over them, and at query time returns peer representations, session context, search hits, or natural-language insights, so you do not have to wire up embeddings and vector retrieval yourself. The core is a FastAPI server you can self-host or use as a managed service, with official Python and TypeScript SDKs, and it works alongside whatever model or orchestration framework you already use. You slot it underneath your existing agent stack as a durable memory layer rather than a ready-made chatbot, and that layer is usually the part teams underestimate when they want agents that stay stateful across sessions.