Hey Product Hunt! 👋 I'm Kim, the maker of Trapic. Here's the problem: every time you start a new AI session, your AI has amnesia. You re-explain your architecture, your conventions, why you chose Postgres over Mongo, why that API is designed that way. Every. Single. Time. I got tired of it. So I built Trapic. What it does: rapic captures your decisions, conventions, and reasoning as atomic knowledge units called "traces." At every session start, your AI auto-recalls he context it needs — no copy-pasting, no re-explaining. What makes it different from just saving notes or using RAG: - Each trace is typed (decision, convention, fact, state) with a causal chain — your AI knows why, not just what - Knowledge has half-lives. A state trace decays in 30 days, a decision in 90, a fact in 365. Stale context gets flagged automatically - When a new decision contradicts an old one, Trapic detects the conflict and links them - It works across every MCP-compatible tool — Cursor, Claude Code, Windsurf, Copilot. One knowledge base, all your AI assistants How to try it: curl -fsSL https://raw.githubusercontent.co... | bash That's it. One line. Works with any MCP tool. We're free during beta and the plugin is open source. I'd love to hear what you think — especially what kind of context you wish your AI remembered. Ship it 🚀

Trapic

Context engineering for AI that actually remembers.

Context engineering for AI that actually remembers.