skills.rest
Atrium-Hermes
Atrium-HermesCOMMUNITY
Explore all skills by this author →
𝕏@atriumhermesGitHub

Autoresearch

Generate four skill variations, score them against a rubric, and commit the winner as a PR.

Updated Jun 2, 2026
Search Tags:#automation#pr#rubric#version-control#skill-improvement#autoresearch#memory-logging
One-click install
npx skills add https://github.com/Atrium-Hermes/atrium-lighthouse --skill autoresearch-atrium-hermes
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Autoresearch
Source: https://github.com/Atrium-Hermes/atrium-lighthouse/tree/main/skills/autoresearch
Command: npx skills add https://github.com/Atrium-Hermes/atrium-lighthouse --skill autoresearch-atrium-hermes

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automatically evolves a target skill by researching improvements, generating variations, and applying the best version.

Core Features & Use Cases

  • Generate four distinct improvements to a target skill and propose them as variations.
  • Score variations against a rubric and commit the winner as a PR into the skill repository.
  • Maintain memory of experiments and log decisions for audit and future iterations.

Quick Start

Set var to a target skill name and run the autoresearch process to generate variations, score them, and apply the winning version.

Frequently Asked Questions about Autoresearch

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

FAQPage Schema
How do I automate skill improvement through iterative research?▼

Automating skill improvement involves researching enhancements, generating four variations, scoring them against a rubric, and committing the winning version as a pull request. This provides versioned updates and maintains a memory log of decisions.

What is the best way to evaluate skill variations before committing?▼

Evaluating skill variations requires scoring each generated version against a predefined rubric. The highest-scoring variation is selected as the winner and committed to the repository as a ready-to-run file.

How do I track decisions when generating skill updates via pull requests?▼

Tracking decisions during skill updates is handled by reading memory logs and writing experiment results to memory files. This maintains an audit trail for future iterations of the target skill.

Can I apply automated research to any skill in my repository?▼

Automated research applies to any skill in the skills directory that benefits from iterative enhancement. The process generates variations, scores them under a rubric, and outputs the result to a new folder.

How does version control work when evolving a target skill automatically?▼

Version control for evolving a skill works by committing the winning variation as a pull request. This ensures all iterative enhancements are tracked and updated via versioned changes in the repository.

What limitations exist when generating skill variations from memory logs?▼

Generating skill variations relies on existing memory logs for context, meaning insufficient historical data may limit enhancement quality. The process produces four variations and requires a rubric to score them effectively.

Related Skills

Explore More Education & Research Skills→
swarm-ai-researchswarm-ai-research

Autoresearch

Generate four skill variations, score them, and commit the winner as a PR.

Official
Advanced
lxt070409lxt070409
1

github-code-review

Automate GitHub pull request reviews using local git and the GitHub API.

Community
Advanced
cfcostacfcosta

autoresearch-create

Set up and run autonomous experiment loops with logging and version control.

Community
Advanced
nklischnklisch
7

skill-evaluator

Evaluate AI agent skills against quality rubrics and generate improvement reports.

Community
Intermediate