Show HN: Moo.md – Mental Models for Claude Code
github.comClaude Code is fast. But fast at what?
moo.md makes it a thinking partner, not just a task executor. Mental models. Confidence gates. Learnings that persist.
When you're stuck on a decision, it runs a pre-mortem. When debugging, Ishikawa. When you need perspective, it channels Rich Hickey or Paul Graham.
Plugins for decisions, writing, and design.
https://github.com/saadshahd/moo.md What I actually get from this: 1. No more "good work" fluff. Claude challenges decisions instead of agreeing. The expert simulations come with confidence ratings and citations — if it's channeling Hickey at 7/10, it tells you why. 2. Compaction anxiety is gone. I used to dread long sessions because insights would disappear when context got too long. Now learnings persist in ~/.claude/learnings/ — patterns from last month are still there. 3. Grounded opinions. When I ask "what would Rich Hickey think?", the response cites his actual talks and documented philosophy. Not hallucinated advice. The mental models aren't the point. The point is Claude stops being agreeable and starts being useful.