# Conversion (#1539): the agentic `flow-test-strategy-execution` skill expressed
# as a declarative YAML Flow. Part of epic #1534. The SKILL.md remains the thin
# wrapper / discoverable trigger surface; this playbook is the source of truth
# for the test orchestration step sequence + the final go/no-go human gate.
#
# Two prose steps dispatch multiple agents in parallel, so they convert to
# `fanout` steps (multi-agent documentation pattern, #1547): Step 4 (E2E + UAT
# panel) and Step 5 (defect triage panel). Step 6 chains report-generating
# agents as a pipeline fanout. Step 7 is the human go/no-go decision (kind: gate).
#
# Schema-fit note (#1539): `inventory`/`targets` omitted — SDLC flows act on a
# single project context rather than iterating a host set.
#
# Faithful to the prose flow (7 prose steps → 7 playbook steps):
#   Step 1 test execution planning   → execution-planning  (test-architect)
#   Step 2 unit suite + coverage      → unit-tests          (test-engineer)
#   Step 3 integration suite          → integration-tests   (test-engineer)
#   Step 4 e2e + UAT (×2 parallel)    → e2e-acceptance      (FANOUT panel + synthesis)
#   Step 5 defect triage + RCA (×2)   → defect-triage       (FANOUT panel + synthesis)
#   Step 6 regression + reports (×3)  → regression-reporting(FANOUT pipeline + synthesis)
#   Step 7 present go/no-go           → present-results     (kind: gate, human decision)
apiVersion: flow.aiwg.io/v1
kind: FlowPlaybook
metadata:
  name: flow-test-strategy-execution
  labels:
    category: sdlc-orchestration
    domain: testing
spec:
  vars:
    test_level: ""
    target_component: ""
    guidance: ""
  steps:
    - id: execution-planning
      capability: teststrat-execution-planning
      inputs:
        - { name: test_level, from: "vars.test_level" }
        - { name: target_component, from: "vars.target_component" }
        - { name: guidance, from: "vars.guidance" }
      outputs:
        - name: test_execution_plan

    - id: unit-tests
      capability: teststrat-unit-tests
      depends_on: [execution-planning]
      outputs:
        - name: unit_results
        - name: coverage_analysis

    - id: integration-tests
      capability: teststrat-integration-tests
      depends_on: [unit-tests]
      outputs:
        - name: integration_results

    # Step 4: E2E execution (qa-engineer) and UAT validation (product-owner)
    # run as an independent panel; results are synthesized into a combined
    # e2e/acceptance result for downstream triage.
    - id: e2e-acceptance
      fanout:
        strategy: parallel
        agents:
          - teststrat-e2e-execution      # qa-engineer
          - teststrat-uat-validation     # product-owner
        synthesize: teststrat-e2e-synthesis   # qa-engineer
      depends_on: [integration-tests]
      outputs:
        - name: e2e_results

    # Step 5: defect triage panel — qa-lead triages all failures, component-owner
    # performs root-cause analysis on P0/P1 defects; synthesized into a triaged
    # defect set.
    - id: defect-triage
      fanout:
        strategy: parallel
        agents:
          - teststrat-defect-triage-panel    # qa-lead
          - teststrat-root-cause-analysis    # component-owner
        synthesize: teststrat-triage-synthesis   # qa-lead
      depends_on: [e2e-acceptance]
      outputs:
        - name: triaged_defects

    # Step 6: regression analysis → comprehensive test report → quality-gate
    # validation chain. Pipeline fanout: each agent consumes the prior output;
    # the project-manager's quality-gate report is the synthesized result.
    - id: regression-reporting
      fanout:
        strategy: pipeline
        agents:
          - teststrat-regression-analysis    # test-architect
          - teststrat-test-report            # test-manager
        synthesize: teststrat-quality-gate   # project-manager
      depends_on: [defect-triage]
      outputs:
        - name: quality_gate_report

    # Step 7: present results + go/no-go. Human reviews the synthesized quality
    # gate decision (PASS | FAIL | CONDITIONAL) and the full test/defect summary.
    - id: present-results
      kind: gate
      description: |
        Human go/no-go gate: present the Test Execution Summary — pass rates by
        level, coverage vs target, defect counts by severity (P0–P3), regression
        status, and the synthesized quality-gate decision (PASS | FAIL |
        CONDITIONAL). The human approves the go/no-go recommendation or returns
        for fixes. The reviewer sees the test-execution-report, coverage-report,
        defect list, regression-report, and test-quality-gate at this pause point.
      depends_on: [regression-reporting]