Guide users through creating, testing, and iterating Claude Skills.
npx skills add https://github.com/zebbern/termstack --skill skill-creator-zebbern
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/zebbern/termstack/tree/main/.github/skills/skill-creator Command: npx skills add https://github.com/zebbern/termstack --skill skill-creator-zebbern
The Skill Creator guides users through the end-to-end lifecycle of building, testing, and iterating Claude Skills. It helps capture intent, author and organize SKILL.md, design evaluation prompts, run automated trigger evaluations, compare results, and apply structured improvements so skills trigger reliably and perform as intended. It also supports iterating descriptions to improve triggering accuracy and packaging finalized skills for deployment.
Describe your intent, draft the skill, run trigger evaluation, and iterate until you are satisfied.
High-intent search queries and answers about installing and using this skill.
To create and refine Claude Skills, use a guided workflow to articulate intent, author SKILL.md, design evaluation prompts, and run automated trigger evaluations. This structured iteration loop applies improvements to reach a robust, trigger-friendly skill ready for deployment.
The best way to evaluate skill triggering accuracy is to generate and run automated trigger-evaluation prompts, collect results, and analyze tradeoffs between configurations. Use eval outcomes to propose description and content improvements for rapid iteration.
You package finalized Claude Skills by validating the iterated SKILL.md content and organizing it into distributable formats. This ensures the skill triggers reliably and performs as intended before sharing or deployment.
Yes, you need pyyaml installed to author and test SKILL.md files, as it is a required dependency for parsing and managing the YAML frontmatter used in skill metadata during the creation and evaluation workflow.
Yes, you can use this meta-skill to compare outcomes between different skill configurations. The evaluation loop collects results and analyzes tradeoffs, enabling you to apply structured improvements based on the compared performance data.
dige04Design and refine prompts and agent personas with structured YAML frontmatter.
JeremiahEllingtonMonitor deployed URLs for console errors, performance regressions, and page failures against baselines.
bacocoAutomate creation, validation, and packaging of Claude skills.
Avi977Automate the end-to-end Claude skill creation loop with SKILL.md and evals.
