Claude Autoresearch Skill — Autonomous goal-directed iteration for Claude Code. Inspired by Karpathy's autoresearch. Modify → Verify → Keep/Discard → Repeat forever.
autoresearch is an agent workflow that repeatedly changes a project, verifies a measurable result, keeps or discards the change, and continues iterating toward a goal. It is for autonomous improvement tasks in Claude Code, OpenCode, and OpenAI Codex across domains with mechanical success measures. The catalogue contains its commands, hooks, skills, plugin, agent, and instruction.
Latest release v2.2.2 — Autoresearch v2.2.2 — Stabilization Release · 12 Aug 2026
30 files for Codex, Claude Code and OpenCode: autoresearch, autoresearch, autoresearch:debug, autoresearch:evals and 26 more — 2,635 tokens loaded in every session.
AGENTS.md A 2,635 tok .agents/skills/autoresearch/SKILL.md A 18 tok .claude/commands/autoresearch.md C 18 tok .claude/commands/autoresearch/debug.md A 21 tok .claude/commands/autoresearch/evals.md A 19 tok .claude/commands/autoresearch/fix.md A 22 tok .claude/commands/autoresearch/improve.md A 16 tok .claude/commands/autoresearch/learn.md C 27 tok .claude/commands/autoresearch/plan.md C 18 tok .claude/commands/autoresearch/predict.md A 13 tok .claude/commands/autoresearch/probe.md A 14 tok .claude/commands/autoresearch/reason.md A 14 tok .claude/commands/autoresearch/regression.md C 34 tok .claude/commands/autoresearch/scenario.md A 16 tok .claude/commands/autoresearch/security.md A 18 tok .claude/commands/autoresearch/ship.md A 20 tok .opencode/commands/autoresearch_debug.md A 20 tok .opencode/commands/autoresearch_evals.md A 18 tok .opencode/commands/autoresearch_fix.md A 21 tok .opencode/commands/autoresearch_improve.md A 15 tok .opencode/commands/autoresearch_learn.md C 27 tok .opencode/commands/autoresearch_plan.md C 17 tok .opencode/commands/autoresearch_predict.md A 12 tok .opencode/commands/autoresearch_probe.md A 13 tok .opencode/commands/autoresearch_reason.md A 13 tok .opencode/commands/autoresearch_regression.md C 34 tok .opencode/commands/autoresearch_scenario.md A 16 tok .opencode/commands/autoresearch_security.md A 17 tok .opencode/commands/autoresearch_ship.md A 19 tok .opencode/agents/docs-manager.md A 20 tok These files are uditgoenka/autoresearch's own configuration — they tell Codex, Claude Code and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.