Hey Product Hunt 👋 Knowl started from a small, daily annoyance: an agent works something out — why we picked this library, what the actual rate limit is, which approach we already tried and rejected — and then the session ends and it's gone. Next session, you paste the explanation again. The obvious fix is "give the agent a memory." That's where the second problem shows up. An append-only memory holds the March answer and the July answer with equal confidence, the agent picks one, and now a wrong answer comes back with a citation attached. Stale knowledge is worse than none, because it gets trusted. So Knowl is built around the change, not the write: → Every fact stores its reasoning, its evidence, and the alternatives it beat. → When a fact changes, the new one supersedes the old. The old one is still there and still queryable — it just stops being the answer. → One knowl init wires up Claude Code, Codex, Cursor, Gemini CLI or Claude Desktop. Same store, all of them. On the first three it also installs lifecycle hooks, so the capture runs by itself. → Local-first. Nothing leaves your machine until you publish it on purpose. Knowl Cloud is the hosted half: one memory your team shares, plus connectors for claude.ai and chatgpt.com so you can ask from the browser. The question I actually want answered: if you run two or more coding agents, what's the thing you're most tired of re-explaining to them?

Knowl

One shared brain. Every AI. Any session. Always current.

One shared brain. Every AI. Any session. Always current.