Hi PH community 👋 I built m8m because I realized my AI memories were everywhere and nowhere. Claude Code had one version of who I am. Codex had another. Cursor had a third. Each tool storing its own memories in its own format, in its own corner of my filesystem. I couldn't see the full picture, I couldn't manage it, and I definitely couldn't take it with me. So I built m8m — one local database that unifies memories across every AI tool you use. You can scan all your providers in one command and see every memory your AI tools have collected. But what I find most useful day-to-day is the MCP server: m8m runs alongside your AI agent, and you can just tell it things — "remember that our staging server is at this IP," "flag that old API key for removal," "what do you know about my AWS setup?" Your agent talks to m8m directly, and every memory gets tracked with full provenance — who wrote it, when, which tool, and a trust score. The security side matters too. Memory poisoning is a real and growing attack vector — a single email can silently rewrite what your AI believes about you. m8m catches credentials, prompt injection patterns, contradictions, and hidden Unicode in your memory files. But security is one lens. The bigger point is: your AI memories are becoming a meaningful part of your digital identity, and you should be able to see, manage, and own them. Everything runs locally. One SQLite file at ~/.m8m/m8m.db — that's your entire memory history. Copy it, back it up, it's yours. I have some ideas for what comes next — think: your memories, not just on this machine. But I'll save that for when it's ready. Try npx @th0t3p/m8m scan and see what your AI tools have been remembering. I'm curious what you find.

m8m

Own your AI memory.

Own your AI memory.