skill-regression-evaluator — Claude Code skill
skill-regression-evaluator is a Claude Code and Codex skill that audits skill versions against regression suites. Catches regressions before knowledge updates break established behavior.
MIT ★ 529 updated 2026-11-17
npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a claude-codeWhat skill-regression-evaluator does
The skill runs test cases against an updated skill definition to catch regressions early. Execution depends on a specified skill version, regression test inputs and expected outputs. The resulting summary flags passed, failed and skipped items.
Unexplained failures block release. When required test data or human inspection is missing, the workflow marks the result as needs-evidence. Output comes as structured evidence, divergence details, suggested fixes and documented boundaries without fabricating scores.
What it helps with
- spot breaking changes in updated skills
- compare skill outputs against expected results
- halt skill releases on unexplained test failures
- track missing test prerequisites with needs-evidence tags
Example prompts for the agent
- Evaluate this skill version against its regression cases and list divergences
- Run regression checks on skill-regression-evaluator before publishing
- Compare the current skill behavior against expected baseline outcomes
How to install skill-regression-evaluator
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a claude-code2Manually: put the files in place
~/.claude/skills/skill-regression-evaluator/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/skill-regression-evaluator/
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a codex2Manually: put the files in place
.agents/skills/skill-regression-evaluator/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a cursor2Manually: put the files in place
.cursor/skills/skill-regression-evaluator/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install ShuaiMXu/happyhands skills/product-design/skill-regression-evaluator2Manually: put the files in place
.github/skills/skill-regression-evaluator/SKILL.md1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a gemini-cli2Platform CLI
gemini skills install https://github.com/ShuaiMXu/happyhands --path skills/product-design/skill-regression-evaluatorManually: put the files in place
.devin/skills/skill-regression-evaluator/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about skill-regression-evaluator
How to install skill-regression-evaluator in Claude Code?
Execute npx skills add ShuaiMXu/happyhands --skill skill-regression-evaluator -a claude-code or save it to ~/.claude/skills/skill-regression-evaluator/SKILL.md.
What does skill-regression-evaluator do?
It executes reproducible regression runs on skill revisions, highlights initial divergences and flags passed, failed and skipped tests.
What is required to run skill-regression-evaluator?
You must supply the target skill version, regression test inputs and expected outcomes.
Source text
SKILL.md · MIT
--- name: skill-regression-evaluator description: "Evaluate a skill version against its regression cases and report first divergences." --- # Skill Regression Evaluator Prevent an expertise update from silently breaking previously accepted behavior. ## Required evidence - skill version - regression inputs - expected outcomes ## Decisions - Run cases reproducibly - Report pass, fail and skipped - Block publication on unexplained failures ## Boundary Passing a finite regression set does not prove universal quality. Return structured evidence, findings, proposed updates and limitations. Use `needs-evidence` when required inputs or professional review are missing. Never invent an Experience Score or promote a project observation into a universal rule.
© ShuaiMXu · MIT · ShuaiMXu/happyhands