uditgoenka/autoresearch
A Claude Code/OpenCode/Codex skill that turns an AI coding assistant into an autonomous goal-directed improvement engine using iterative verify-keep-discard loops.
Velocity · 7d
+6.4
★ / day
Trend
→steady
star history
This skill wraps an AI coding assistant (Claude Code, OpenCode, or OpenAI Codex) in an autonomous iteration loop. The user sets a goal and a mechanical metric; the agent then repeatedly modifies code, verifies against the metric, and keeps or discards changes. It supports 13 commands and 9 safety hooks to guide the iteration process with guardrails. The approach is inspired by Karpathy’s autoresearch methodology.
Frequently asked
- What is uditgoenka/autoresearch?
- A Claude Code/OpenCode/Codex skill that turns an AI coding assistant into an autonomous goal-directed improvement engine using iterative verify-keep-discard loops.
- Is autoresearch open source?
- Yes — uditgoenka/autoresearch is open source, released under the MIT license.
- What language is autoresearch written in?
- uditgoenka/autoresearch is primarily written in Shell.
- How popular is autoresearch?
- uditgoenka/autoresearch has 5.4k stars on GitHub and is currently holding steady.
- Where can I find autoresearch?
- uditgoenka/autoresearch is on GitHub at https://github.com/uditgoenka/autoresearch.