autoresearch

Name: autoresearch
Availability: InStock
Author: air-gapped

Execute autonomous experiments to improve software artifacts using a measurable metric.

5|1|Updated Apr 19, 2026

Search Tags:#git #metrics #agent #provenance #experiments #loop #autoresearch

Ready to run

npx skills add https://github.com/air-gapped/skills --skill autoresearch-air-gapped

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/air-gapped/skills/tree/main/.claude/skills/autoresearch
Command: npx skills add https://github.com/air-gapped/skills --skill autoresearch-air-gapped

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Karpathy-pattern autoresearch enables autonomous iteration to improve code, models, or prompts by running measured experiments and tracking provenance over time.

Core Features & Use Cases

Supports three modes: Optimize, Research, and Improve to cover numeric optimization, deep multi-angle analysis, and research-informed improvements.
Provides a structured experiment loop including a Truth Layer, Mutable Surface, Verifier, and Metric, plus automatic baseline establishment and a results ledger for replayability.
Includes references for deeper guidance and established patterns such as STORM multi-perspective research and blind validation to mitigate bias.

Quick Start

Start the autoresearch loop by selecting a mode and issuing the command, for example /autoresearch optimize to start optimizing.

Frequently Asked Questions about autoresearch

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

FAQPage Schema

How do I run autonomous experiments to improve code using measurable metrics?▼

To run autonomous experiments, you select a mode like Optimize, define a mutable surface, a truth layer for tests, a numeric verifier, and a metric direction to iterate and track provenance over time.

What is an autonomous research loop for AI prompts and configurations?▼

An autonomous research loop iteratively modifies prompts or configurations, measures results against a numeric verifier, and logs the experiments to establish baselines and track provenance for replayable improvements.

Do I need tests or benchmarks to start autonomous code optimization?▼

Yes, you need a truth layer consisting of tests or benchmarks to serve as the numeric verifier, ensuring that each iteration of the mutable surface is measured against a clearly defined metric and direction.

What is the best way to track provenance when iterating on software artifacts?▼

The best way to track provenance is using a strict branch and log workflow that records each experiment in a results ledger, establishing baselines and ensuring the entire optimization process remains replayable.

Can I use multi-perspective research to inform software artifact improvements?▼

Yes, the Research mode utilizes multi-perspective research patterns like STORM to perform deep analysis, informing subsequent modifications to the mutable surface before validating them with the numeric verifier.

When should I not use an autonomous research loop for artifact optimization?▼

You should avoid this loop if your software artifact lacks a clearly defined mutable surface, a measurable numeric verifier, or a specific metric direction, as these are strictly required for the experiment workflow.