Show HN: Sibyl – self-hosted cross-agent memory for AI coding agents

I realized last year that I needed a better shared substrate for all the parallel coding (and other) agents I run if I wanted all my ADHD agentically-engineered dreams to come true. It started off as a simple Kanban board for agents with a crawler and rag, but it's long since evolved into a high performance, scalable, multi-user, self-hostable agent memory system built on SurrealDB that I can't live without. It's the thing that's helped me go absurdly parallel without just turning into a slop cannon.

If an agent can use a CLI (or MCP, but CLI works better) it can speak Sibyl. You can run it locally, on a VPS in Docker, on Kubernetes, or wherever. The benchmarks and evals are competitive (96.96% strict recall@5 on LongMemEval-S, on the live API path, with no LLM in the retrieval path), and are actively expanding.

My Sibyl knows everything about all of my projects, and it's ingested all my notes, docs, and even email. I can start an agent anywhere and it immediately knows where to go next. And it's all stored on my own infra.

I just tagged 1.0.2 and use it every day. It's Apache 2.0 licensed and totally open source, contributions are welcome. Check it out and tell me if it works for you as well as it does for me!

https://github.com/hyperb1iss/sibyl