Test and verify
Skill sainzs/santiagosainz-skills/skills/core/test-and-verify
Auditable, portable agent skills that turn open-source maintenance - review, debugging, verification, handoff - into repeatable workflows for Codex, OpenCode, and Pi
npx -y skills add sainzs/santiagosainz-skills --skill test-and-verifyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Test and verify one behavior through a tight evidence loop: baseline, failing proof when useful, minimal edit, and final proof. Use for tests, verification, TDD, regression-proofing, smoke checks, readiness validation, or prove-change.
SKILL.md
1.8 KB, as published. Nobody here has run it
test-and-verify
Ship one behavior with evidence.
Unix/Pi Contract
- Do one job: prove one behavior changed or still works.
- Consume a target behavior and repo commands; emit changed files plus exact proof.
- Prefer the repo-native runner and the narrowest meaningful command.
- Act, then verify. Do not narrate every tool step.
- After two identical failures, change approach instead of looping.
Use
Use when changing behavior, fixing a bug with a known expectation, hardening a workflow, or proving an app can actually run.
Proof Ladder
Choose the first proof that gives real confidence:
- Existing failing test or check.
- New behavior test through a public interface.
- CLI/API/browser smoke repro.
- Build/type/lint gate for structural changes.
- Manual evidence only when automation is impractical.
Workflow
- Establish the baseline with the narrowest meaningful command.
- Add a failing proof first when the behavior is testable and risk justifies it.
- Make the smallest edit that satisfies the proof.
- Refactor only while green.
- Run the original proof and any adjacent gate needed for confidence.
- Report exact commands and outcomes.
Test Quality
- Test behavior through public interfaces.
- Avoid mocks of internals unless the boundary is external, slow, costly, or nondeterministic.
- Do not bulk-write imagined tests ahead of implementation.
- Prefer one useful regression over many brittle assertions.
Output
Return:
- Behavior proved.
- Files changed.
- Proof added or reused.
- Commands run and results.
- Residual risk.