| name | fabric-test |
| description | The QA skill — reviews architecture for testability, creates test plans from acceptance criteria, and validates deployments. Use when user says "create test plan", "validate deployment", "run validation", "review architecture", "check items", "what should we test", or after architecture is finalized or deployment is complete. Do NOT use for architecture design (use fabric-design) or deployment (use fabric-deploy).
|
| pre-compute | ["test-plan-prefill"] |
Fabric Testing & Validation (QA Role)
Mode 1: Architecture Review + Test Plan (Phase 2a)
⚡ Fast-forward: This phase is auto-completed by run-pipeline.py fast-forward. The test plan is pre-filled by test-plan-prefill.py. This skill is only invoked if fast-forward fails or the user requests manual review.
Step 1: Review Pre-Filled Test Plan
The test plan at _projects/[name]/docs/test-plan.md is pre-generated. Review for:
- Testability concerns (
red blockers, yellow warnings)
- Missing edge cases or expected results
Enhance if needed; do not rewrite from scratch.
Set review_outcome: approved (no red) or concerns (has red).
Mode 2: Post-Deployment Validation (Phase 3)
For live workspace validation:
python .github/skills/fabric-test/scripts/validate-items.py _projects/[name]/docs/deployment-handoff.md
Follow the checklist from tool output. Write _projects/[name]/docs/validation-report.md with status: passed | partial | failed.
Constraints
- Never invent ACs not in the Architecture Handoff
- Never modify Fabric items — validate only
Handoff
After producing the output file, advance:
python _shared/scripts/run-pipeline.py advance --project <project-name> -q
If the output shows 🟢 AUTO-CHAIN → <skill>, invoke that skill immediately — do NOT stop and ask the user.
Only 🛑 HUMAN GATE (Phase 2b sign-off) requires user action.