leo-lilinxiao /
codex-autoresearch
Codex Autoresearch Skill — A self-directed iterative system for Codex that continuously cycles through: modify, verify, retain or discard, and repeat indefinitely. Inspired by Karpathy’s autoresearch concept.
Updated 8 min ago
+69 stars gained this week
Stars gained each day, last 30 days
Usual pace · 12 / day
Oct 9Oct 24
This week Before Usual pace
In short
A Codex skill that runs measurable, iterative experiments: it changes code, verifies a command-based metric, keeps improvements and reverts failures until a target is reached. For developers wanting autonomous optimization of tests, coverage, latency and similar outcomes.
Summary written by AI from the repository’s description and README.
What this repo contains
- Its README
- Suggests turning off the agent's permission prompts
- Skills
- 1 SKILL.md file
- Scripts
- 12 in skill folders Python 9 · Shell 3
No executables, hook files or MCP settings.
Read from the repository's files and README: facts to check on GitHub, not a review.
leo-lilinxiao / codex-autoresearch
View on GitHub(opens in new tab)