Autoresearch for skills

Install

git clone https://github.com/olelehmann1337/autoresearch-skill ~/.claude/skills/autoresearch

This is third-party code. Review the repository files before installing.

What it does

The skill adapts Andrej Karpathy's autoresearch approach from model training to Claude Code skill prompts. It first collects the target SKILL.md path, 3-5 test inputs and 3-6 yes or no checks from the user. It then loops: runs the skill, scores every output, applies one prompt mutation and keeps it only if the score improves. The output is an improved SKILL.md, a results.tsv log, a changelog of every attempt and an HTML dashboard to watch progress.

Who it is for. For skill authors whose skill works inconsistently and who want measurable quality gains.

Example request

Run autoresearch on my skill at ~/.claude/skills/newsletter/SKILL.md with a 10-cycle budget

Limitations

There is no README; all documentation is in SKILL.md and eval-guide.md. By default the loop has no iteration cap and uses many model calls, so set a budget first. SKILL.md links to references/eval-guide.md while the file sits at the repository root, so move it when installing.

How to disable. Delete the autoresearch folder from ~/.claude/skills/.

Security check

• Runs another skill repeatedly and rewrites its SKILL.md
• Can burn many tokens without a budget cap

README in short

The repository has no README. SKILL.md covers collecting context before starting, building the eval suite, a baseline run, the mutation loop and the output file format. eval-guide.md explains how to write good binary checks and which mistakes to avoid.

SKILL.md

---
name: autoresearch
description: "Autonomously optimize any Claude Code skill by running it repeatedly, scoring outputs against binary evals, mutating the prompt, and keeping improvements. Based on Karpathy's autoresearch methodology. Use when: optimize this skill, improve this skill, run autoresearch on, make this skill better, self-improve skill, benchmark skill, eval my skill, run evals on. Outputs: an improved SKILL.md, a results log, and a changelog of every mutation tried."
---

# Autoresearch for Skills

Most skills work about 70% of the time. The other 30% you get garbage. The fix isn't to rewrite the skill from scratch. It's to let an agent run it dozens of times, score every output, and tighten the prompt until that 30% disappears.

This skill adapts Andrej Karpathy's autoresearch methodology (autonomous experimentation loops) to Claude Code skills. Instead of optimizing ML training code, we optimize skill prompts.

---

## the core job

Take any existing skill, define what "good output" looks like as binary yes/no checks, then run an autonomous loop that:

1. Generates outputs from the skill using test inputs
2. Scores every output against the eval criteria
3. Mutates the skill prompt to fix failures
4. Keeps mutations that improve the score, discards the rest
5. Repeats until the score ceiling is hit or the user stops it

**Output:** An improved SKILL.md + `results.tsv` log + `changelog.md` of every mutation attempted + a live HTML dashboard you can watch in your browser.

Related