Convert user intent into a trigger-optimized SKILL.md with evaluation workflows.
npx skills add https://github.com/tnunamak/dotfiles --skill skill-creator-tnunamak
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/tnunamak/dotfiles/tree/main/ai/skills/local/skill-creator Command: npx skills add https://github.com/tnunamak/dotfiles --skill skill-creator-tnunamak
Creating a new Claude skill—or improving an existing one—can be guesswork because you need good triggering language and you must verify that the skill actually performs the intended workflow.
Tell the AI what you want the skill to do and when it should trigger, and ask it to create the initial SKILL.md plus 2–3 test prompts for you to run.
High-intent search queries and answers about installing and using this skill.
To create a Claude skill, convert your intended goal into a trigger-optimized SKILL.md file containing a clear description and operational instructions. This process ensures the skill activates reliably when needed and avoids misfiring.
Evaluation-driven iteration designs test prompts, runs them against your skill, and grades the outputs using quantitative pass/fail metrics. This workflow aggregates results with variance-aware benchmarking to measure improvements accurately.
To fix over-triggering or under-triggering, run trigger accuracy evaluations using generated test prompts. The eval results provide evidence to update the SKILL.md description, optimizing when the skill should or should not activate.
Yes, you can benchmark Claude skill performance using quantitative pass/fail metrics and variance-aware analysis. This approach grades expectations with evidence across multiple iterations until the skill meets your satisfaction.
No external dependencies are required to build and test Claude skills. The workflow operates by generating a SKILL.md file and evaluating test prompts directly within the Claude environment.
To start drafting a new skill, tell the AI what you want the skill to do and when it should trigger. It will generate an initial SKILL.md along with two to three test prompts for you to run and evaluate.
daikichiba9511Automate the creation, evaluation, and iteration of Claude skills with SKILL.md.
C0ntr0lledCha0sImprove Claude Code component descriptions, permissions, triggers, and documentation.
MisLinkDraft SKILL.md files, generate test prompts, and run with-skill versus baseline evaluations.
zhbonerDraft SKILL.md files and refine them through prompt-based test evaluations.
