Spine position: Between slice-tasks and plan-work for epics with risk: P0|P1. Optional (can be waived in state.yaml) for P2/P3.

Bridges the gap between slicing and planning by systematically designing the test suite as a first-class system. Produces specs/tech-architecture/eNN-TEST_PLAN_LATEST.md.

• Read specs/epics/eNN-*/ story list.

1. Analyze Epic: Read sliced stories in the active epic capsule.
2. Risk Assessment: Map behaviors to P0–P3 risk tiers.
3. Level Strategy: Classify each scenario (Unit, Integration, E2E).
4. Fixture Design: Plan factories, network intercepts, mocks (see REFERENCE.md).
5. NFR Plan: Define verifiable commands for Non-Functional Requirements. Skip if --lite.
6. Publish: Generate specs/tech-architecture/eNN-TEST_PLAN_LATEST.md.

• DO NOT write test code or production code during this skill.
• Scenario ID format MUST be: SC-eNNsYY-P{0|1|2|3}-NN.
• plan-work MUST reference these scenario IDs in its §17 Gherkin acceptance criteria.
• Default to pushing tests to the lowest possible level.

• $bmad plan-tests: Standard execution.
• $bmad plan-tests --lite: Skips NFRs and complex fixture planning.

→ verify: ls specs/tech-architecture/*-TEST_PLAN_LATEST.md 2>/dev/null | head -1 && echo OK || echo "SKIP: no test plan yet — EPIC not set or not generated"

Gate: READY -> next: plan-work Writes: state.yaml handoff.next_skill = plan-work