Source checked · Research
Autoresearch
What the publisher saysa Codex, Claude Code, and OpenCode skill for bounded, metric-driven loops that modify files and keep or revert each experiment.
Best for: Source discovery for teams evaluating how to run a bounded improvement experiment on a disposable branch with a measurable goal.
Limitations: Unlimited loops, production actions, broad repository writes, or experiments without a reviewed metric, verification command, and stop condition.
- Public use
- 5,321 GitHub stars · 394 forks
- Activity
- Release v2.2.1 on 23 Jun 2026 · repository updated 23 Jun 2026
- Community
- 3 open issues · Discussions available
- Checked
- 14 Jul 2026
Source facts only. Not tested or recommended by Agent Blocks.