skills.rest
wjgoarxiv
wjgoarxivCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

autoresearch-skill

Automate hypothesis generation, experimentation, and evaluation across LLM CLI platforms.

29|4|Updated Mar 15, 2026
Search Tags:#experiment automation#autonomous research#iterative optimization#LLM CLI#mechanical evaluation
One-click install
npx skills add https://github.com/wjgoarxiv/autoresearch-skill --skill autoresearch-skill
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: autoresearch-skill
Source: https://github.com/wjgoarxiv/autoresearch-skill/tree/main
Command: npx skills add https://github.com/wjgoarxiv/autoresearch-skill --skill autoresearch-skill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill automates the process of experimental research, hypothesis generation, and iterative improvement using an AI agent. It streamlines the research process across various domains and applications.

Core Features & Use Cases

  • Autonomous Research Loop: Automates the 5-stage process of understanding, hypothesizing, experimenting, evaluating, and logging to iterate on research goals.
  • Multi-Agent Compatibility: Supports integration with Claude Code, Codex CLI, OpenCode, and Gemini CLI.
  • Mechanical Evaluation: Ensures objective, mechanical evaluation of outcomes using a JSON contract format.
  • Overnight Persistence: Continues research overnight or across sessions.
  • Use Cases: Ideal for optimizing code, fine-tuning ML models, exploring literature, and auditing systems.

Quick Start

Start an autonomous research loop for optimizing a sorting algorithm by running: /autoresearch

Frequently Asked Questions about autoresearch-skill

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

FAQPage Schema
How do I automate the iterative research process for hypothesis generation and experimentation?▼

Automate iterative research by running a 5-stage autonomous loop that handles understanding, hypothesizing, experimenting, evaluating, and logging. Use the /autoresearch command to start optimizing code or fine-tuning ML models automatically.

What is mechanical evaluation in autonomous research loops?▼

Mechanical evaluation ensures objective assessment of experimental outcomes using a JSON contract format. This enforces structured, repeatable evaluation criteria for autonomous research iterations across different domains and platforms.

Can I run autonomous experiment automation overnight across multiple LLM CLI platforms?▼

Yes, autonomous experiment automation supports overnight persistence and integrates with Claude Code, Codex CLI, OpenCode, and Gemini CLI. It continues the research loop across sessions without requiring continuous manual intervention.

Do I need specific dependencies to run autonomous research loops?▼

You need Python 3.8+ standard library and specific LLM CLI tools installed for execution. No additional external dependencies are required to run the autonomous research loop.

What's the best way to evaluate experimental outcomes objectively in automated research?▼

The best way to evaluate experimental outcomes objectively is using a JSON contract format for mechanical evaluation. This ensures the autonomous research loop produces consistent, unbiased results during iterative optimization.

Related Skills

Explore More Software Engineering Skills→
BumblebiberBumblebiber
23

hmem-setup

Install and configure the hmem MCP server across system-wide or project-local scopes.

Community
Advanced
slowman2084slowman2084
6

meta-log-converter

Convert platform test stdout logs into ShareGPT-format JSON files.

Community
Intermediate
EndUser123EndUser123

ai-pcli

Orchestrate parallel LLM CLI queries across qwen, gemini, codex, and opencode backends.

Community
Advanced
AlexiosBluffMaraAlexiosBluffMara
2

ml-intern

Automate ML post-training research, code generation, and experiment orchestration.

Community
Advanced