Autoresearch Pro — Development skill for Claude Code
Improve existing agent skills, prompts, articles and workflows through bounded, reversible experiments.
How to install Autoresearch Pro
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open 0xcjl/autoresearch-pro and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Autoresearch Pro does
Improve existing agent skills, prompts, articles and workflows through bounded, reversible experiments. Compatible with Codex, Claude Code, Hermes and OpenClaw.
Alternatives in Development
- Contributing — Thanks for your interest in contributing to Marketing Skills 13.8k ★
- Hermes Agent Self Evolution — ⚒ Evolutionary self-improvement for Hermes Agent — optimize skills, prompts, and code using DSPy + GEPA 5.2k ★
- Autoresearch — Claude Autoresearch Skill — Autonomous goal-directed iteration for Claude Code 2.5k ★
README
autoresearch-pro
Improve an existing skill, prompt, article, plugin, workflow, or system mechanism through small, reversible changes and before/after comparisons.
[中文说明](README_zh.md) · [Core instructions](SKILL.md)
The skill was renamed from **cjl-autoresearch-cc** to **autoresearch-pro**. This repository was also renamed to autoresearch-pro. The folder and frontmatter name used by your agent should be `autoresearch-pro`.
When to use it
Use it for an explicit iteration, a case-based comparison, or an improvement grounded in failure records. It is not needed for a one-off rewrite, translation, an explanation of autoresearch, or routine development without an experiment.
Example request:
Use autoresearch-pro to improve this classification prompt. Preserve the original, compare both versions on a few realistic inputs, and deliver one useful revision. Do not use paid APIs or modify external systems.
The default is **one round**, with at most **three** when a concrete remaining problem justifies another round. Save the baseline, test the same tasks, inspect actual artifacts, and keep only demonstrated improvements. If runtime testing is unavailable, deliver a separate candidate with that limitation instead of claiming a quality gain. No mandatory generic checklist, simulated quality percentage, or 100-round loop.
Install
No package manager dependencies, scripts, credentials, or provider configuration are required. Clone to a stable directory named after the skill:
mkdir -p ~/.agents/skills
git clone https://github.com/0xcjl/autoresearch-pro.git ~/.agents/skills/autoresearch-pro
These commands assume the destination does not exist. If you already have an installation, inspect and back it up first; do not overwrite local edits. Record `git rev-parse HEAD` in the checkout for reproducibility. To update a clean checkout later, use `git pull --ff-only`; review the diff before reloading clients.
Agent-specific setup
| Agent | Entry po
Related Skills
Debug Pro
Resolve existing defects through adaptive evidence gathering, exact reproduction, causal confirmation, bounded
Skill Autoimprove
OpenClaw skill auto-improvement system using Karpathy autoresearch loop
Compare Experiments
Check if two ENCODE experiments are compatible for combined analysis
Improve Extension
Improve a Chrome extension using a measured, independent, bounded candidate-selection loop.
Plugin Improve
Run one bounded, non-regressing improvement pass on a plugin
Polish Content
Polish existing blog articles to publishing-ready quality. Fixes structure, readability, CTAs, and scoring. Us
Related Agents
Autoresearch Orchestrator
Use this agent to run an autoresearch experiment session end-to-end — setup, baseline, and a batch of edit→mea
Autoresearcher
One-bounded-dig research experimenter — reads the brief, proposes ONE hypothesis targeting ONE harness compone
Ensemble Curator
High-volume triage agent for external-AI-agent PRs (Codex, Jules, Hermes, Droid, Aider, etc.). Analyzes, conso