| name | verification-ladder |
| description | Use this skill when deciding how deeply to validate a change or when explaining confidence in a code change. Trigger for test planning, selecting the right validation rung, or classifying confidence as test-green, smoke-green, or watched-live-green. Do not use this skill for runtime root-cause investigation or slice close-out unless validation choice is the main task. |
Verification Ladder
Scope: General.
Use this skill when the main question is: how much validation is required, or what confidence level is honest?
This skill owns:
- selecting validation depth
- mapping change type to test/integration/smoke/watched-run expectations
- describing confidence honestly
This skill does not own:
- runtime root-cause debugging
- proposal hygiene
- slice commit/push closure
Use $runtime-debugger when the issue is a live runtime failure.
Use slice-closeout when the main job is closing an implementation slice.
Ladder
- Crate or unit tests
- Integration or e2e tests
- Binary smokes
- Watched live run
Decision Rules
Climb the ladder when the change affects IPC, networking, concurrency, supervision, routing, delivery, materialization, or environment-specific behavior.
Output Expectations
Report:
- which rungs were run
- what passed
- what remains unverified
- the highest honest confidence level
