A Claude Code mod that reverts its own failed ideas
Try an idea, benchmark it, keep the win, roll back the regression, repeat. A community mod runs that loop inside Claude Code. It's an almost 1:1 port of pi-autoresearch and uses the new mods API to reshape the UI.
Sessions use the same .auto/log.jsonl format as pi-autoresearch, so you can swap logs between the two tools. The project lists karpathy/autoresearch as its inspiration and says the port also fixes a few small bugs.
For a toy demo, copy examples/sort-bench into a git repo and run /autoresearch make sort.js faster. You'll need Claude Code 2.1.285 or newer, plus CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1 in ~/.claude/settings.json.
Would you let a loop decide which of your changes survive?
Related stories
- Claude Code ships mods, the same tool behind its /diff
- Claude Code shipped its own TypeScript in npm source maps
- Show HN: Generate a Claude Code or Codex agent fleet from one HTML file
- pairmark races Claude Code against Codex in your repo
- Oh My Subagents keeps Claude Code teams running
- Show HN: Agent-hop – a Rust TUI that hops a live Claude Code chat into Codex
Comments
No comments yet. Be the first.
Join the conversation
Sign in with Google to leave a comment. Your name and avatar come from your Google profile, and the comment appears after moderation.