probe-definelisted
Install: claude install-skill opendatadiscovery/odd-team
# Generate a probe skeleton from an uncovered cell (slice 4)
Drive the layer-5 probe-corpus growth. Each invocation produces (or refreshes) a probe-skeleton YAML at `lineage/{repo}/probes/{P-NNN}.yaml`. The skeleton carries enough scaffolding for the maintainer to fill in the arrange / act / observe / assert blocks with the specific behaviour the probe pins; the cross-references + frontmatter are pre-populated from feature-flows.yaml.
This skill is the maintainer-facing entry point for the canonical slice-4 deliverable per `adrs/drafts/dynamic-verification-layer.md`.
## Prerequisite
- `lineage/{repo}/feature-flows.yaml` exists with at least one feature.
- `lineage/{repo}/probes/` directory exists (`/probe-run` created it as part of slice 2).
- `python3` available (the skill uses a small inline script to compute the next-available probe ID and parse feature-flows.yaml).
## Argument forms
| Form | Behaviour |
|---|---|
| `/probe-define --feature <feature-id>` | Generate one skeleton for the first uncovered cell of that feature (lowest test_class index in `unit/integration/performance/security` order). |
| `/probe-define --feature <feature-id> --test-class <class>` | Generate one skeleton for the specified test_class of that feature. |
| `/probe-define --feature <feature-id> --all-uncovered` | Generate one skeleton per uncovered cell of the feature (up to 4 — one per test_class). |
| `/probe-define --feature <feature-id> --out-id P-NNN` | Override the auto-derived probe ID