MCP Memory Gateway
Thumbs down a mistake. Your AI agent never repeats it.
2 followers
Thumbs down a mistake. Your AI agent never repeats it.
2 followers
ThumbGate is the safety net for AI coding agents. Give a thumbs-down to a bad action once, and your agent will never ship that mistake again. Every downvote becomes a concrete prevention rule and a pre-action gate that blocks known-bad actions before they execute across 36 tools (git, shells, files, HTTP, clouds). v0.8.4 ships improved rule validation, faster gate checks, and a live dashboard. Unlike Mem0 (memory) or SpecLock (specs), ThumbGate enforces lessons at the moment of action.
I built ThumbGate after watching "smart" coding agents repeat the same destructive mistakes: force-pushing to main, nuking uncommitted work, or curling into production without a safety net. One thumbs-down from me should have been enough — but the agent kept doing it because feedback never turned into hard constraints.
ThumbGate is my answer: every 👎 becomes a prevention rule and a pre-action gate that blocks the agent before it can hurt your repo, terminal, or infra again. In ~10 seconds you can point it at your existing Claude, Cursor, or MCP-compatible agent and get a live dashboard showing what it's trying to do, which rules fired, and what was blocked.
This isn't RLHF or model finetuning — it's a brutally honest enforcement layer. Compared to Mem0 (great for long-term memory/retrieval), ThumbGate is opinionated about one thing: "you saw this go wrong once, it must never pass through again." Compared to SpecLock (great for locking detailed PRDs), ThumbGate is for vibe-coding reality where you only notice the problem after GitHub or your cloud bill takes the hit.
Honest disclaimer: v0.8.4 is battle-tested on my own agents and early users but still evolving — treat it as a strong safety net, not a license to stop reviewing diffs. Would love your feedback on what to enforce next.