ResearcherSkill
Autonomous experiment loops that optimize any measurable metric
All Skills in This Repository (1)
Pure Emerald Level IndicatorsFrequently Asked Questions
FAQPage SchemaHow to install ResearcherSkill?โผ
Run `npx skills add krzysztofdudek/ResearcherSkill --all -g -y` in your terminal to install the skill globally for your agent.
What problems does the researcher skill solve?โผ
It optimizes anything measurable, such as slow builds, flaky tests, API latency, or prompt accuracy, by running repeated experiments and keeping only changes that improve your metric.
How does the experiment loop work?โผ
The agent interviews you about the goal and metric, sets up a lab on a git branch, then commits, measures, and logs each experiment autonomously, discarding failures and forking new strategies when stuck.
Does it work with Claude Code, Codex, and Cursor?โผ
Yes. The single SKILL.md file follows the universal skill standard and runs in Claude Code, GitHub Copilot CLI, Codex, Cursor, and any agent that reads markdown skills.
Can it improve things without a numeric metric?โผ
Yes. For qualitative goals like prompt quality, it builds a scoring rubric with you and uses multiple independent evaluator subagents to score each experiment fairly.
Related Repositories in Education & Research
View All in Education & Researchโlearn-claude-code
Learn to build AI agent harnesses from scratch
ai-agents-for-beginners
Learn to build, deploy, and secure AI agents step by step
33-js-concepts
Master 33 essential JavaScript concepts with tested, verified guides