skills.rest
cdeistopened
cdeistopenedCOMMUNITY
Explore all skills by this author →
GitHub

autoresearch

Run iterative prompt mutation experiments on Claude Code skills and generate updated SKILL.md files.

2|1|Updated Mar 18, 2026
Search Tags:#workflow#evaluation#experimentation#prompt-optimization#skill-improvement#autoresearch
One-click install
npx skills add https://github.com/cdeistopened/skill-stack-skills --skill autoresearch-cdeistopened
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/cdeistopened/skill-stack-skills/tree/main/plugin-meta/autoresearch
Command: npx skills add https://github.com/cdeistopened/skill-stack-skills --skill autoresearch-cdeistopened

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates autonomous evaluation and iterative improvement of Claude Code skills by running multipath tests, scoring outputs with binary checks, and mutating prompts to drive higher quality results.

Core Features & Use Cases

  • Autonomous experimentation loops inspired by Karpathy's autoresearch methodology for Claude Code skills.
  • Binary-eval scoring, controlled mutations, and changelog generation to track progress.
  • Produces updated SKILL.md copies, results logs, and a live dashboard to monitor progress across runs.

Quick Start

Initiate autoresearch on a target skill by supplying the path to its SKILL.md and starting the loop to generate an improved skill version along with its results and changelog.

Frequently Asked Questions about autoresearch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate Claude Code skill prompt optimization?▼

Automate Claude Code skill prompt optimization by running iterative experiments with binary evaluators to score outputs, mutate prompts, and generate an updated SKILL.md with results logs and a changelog.

What is autonomous prompt mutation for skill improvement?▼

Autonomous prompt mutation is a methodology that runs deterministic mutation cycles on Claude Code skills, applying binary checks to evaluate and iteratively drive higher quality prompt outputs.

Can I benchmark Claude Code skills across multiple evaluation scenarios?▼

Yes, you can benchmark Claude Code skills across multiple evaluation scenarios by supplying the path to a SKILL.md and initiating experimentation loops that track progress and produce a live dashboard.

Does iterative prompt experimentation preserve the original Claude Code skill?▼

Iterative prompt experimentation preserves the original Claude Code skill by generating updated SKILL.md copies, results logs, and changelogs separately rather than overwriting the initial file.

How do I track changelog progress during Claude Code prompt evaluation?▼

Track changelog progress during Claude Code prompt evaluation by running autonomous experimentation loops that log binary scoring results and mutations, producing a live dashboard to monitor runs.

Related Skills

Explore More Software Engineering Skills→
akillnessakillness
40

skill-autoresearch

Optimize existing skills by scoring test outputs with binary evals.

Community
Advanced
athan-dialathan-dial

folio:export

Compile LaTeX documents into PDFs and package final exports.

Community
Intermediate
DoeringChristianDoeringChristian

skill-writer

Guides creation of Agent Skills for Claude Code with SKILL.md files.

Community
Intermediate
rattamnoonrattamnoon

skill-creator

Build, evaluate, and refine Claude Skills with automated evaluations.

Community
Advanced