Simulate trigger decisions to evaluate skill descriptions with pass/fail results.
npx skills add https://github.com/danmestas/wardrobe --skill skill-eval-runner
Please help me install this Agent Skill. Skill: skill-eval-runner Source: https://github.com/danmestas/wardrobe/tree/main/skills/skill-eval-runner Command: npx skills add https://github.com/danmestas/wardrobe --skill skill-eval-runner
This tool automates binary pass/fail evaluations of skill descriptions by simulating trigger decisions, enabling quick verification of whether a skill would be invoked by a given prompt without external LLM judgments.
Invoke /eval skill <name> to run the binary trigger tests for that skill.
High-intent search queries and answers about installing and using this skill.
You can automate binary skill evaluations by simulating trigger decisions against skill descriptions, running strict pass/fail checks to verify whether a prompt invokes a skill without external LLM judgments.
To verify prompt routing after edits, the evaluation runner automatically re-evaluates triggers whenever a skills/*/SKILL.md file changes, ensuring your trigger alignment remains accurate and up-to-date.
Yes, you can run prompt regression checks locally without external API calls by using built-in heuristics that read skill frontmatter and eval specs to judge description-trigger alignment and report pass/fail outcomes.
The best way to test prompt triggering locally is to simulate trigger decisions against the skill frontmatter description, producing per-prompt results and a final pass/fail outcome based on alignment heuristics.
No external dependencies or components are required to run skill validation checks; the runner uses local reasoning and built-in heuristics to read frontmatter directly and evaluate trigger alignment.
prthikList, add, remove, edit, and search local skill definitions.
SethGammonExecute Write, Edit, Bash, and Read calls to test Citadel hook pipeline health.
KYRIE66nbExplain routing decisions for user requests with structured decision overviews.
knowledgeislandsAudit Claude Code subagent definitions against mechanical and judgment criteria.
