qa-workflow-cc 1.0.0

package/skills/qa/workflows/report-phase.md

@@ -0,0 +1,64 @@
+# Workflow: Report Phase (Phase 4)
+
+<purpose>
+Consolidate raw test results into a single QA report using the report writer agent.
+</purpose>
+
+<required_context>
+- PROFILE: .claude/qa-profile.json (project name, apps)
+- STATE: docs/qa-reports/cycle-state.json (cycle number, rawResultsDir)
+- MODEL_MAP: resolved model for qa-report-writer
+- RAW_RESULTS: docs/qa-reports/cycle-{N}-raw/*.md
+</required_context>
+
+<process>
+
+## 1. Read Raw Results
+
+Read all files from `docs/qa-reports/cycle-{N}-raw/`:
+- List files in directory
+- Read each file to get agent results
+
+## 2. Identify Previous Cycle (for trends)
+
+If cycle > 1, locate previous cycle report:
+`docs/qa-reports/cycle-{N-1}-*.md`
+
+## 3. Spawn Report Writer
+
+Inline raw results paths and previous cycle path into the prompt:
+
+```
+Task(
+  subagent_type="general-purpose",
+  prompt="[QA-REPORT-WRITER] Read .claude/agents/qa-report-writer.md.
+  IMPORTANT: Read the Resources section FIRST.
+
+  PROJECT CONTEXT (inlined from profile):
+  - Project: {PROFILE.project.name}
+  - Apps: {JSON summary of PROFILE.apps}
+  - Cycle: {N}
+
+  RAW RESULTS TO CONSOLIDATE:
+  {list of file paths in cycle-{N}-raw/}
+
+  Write consolidated report to: docs/qa-reports/cycle-{N}-{date}.md
+  Previous cycle report: {path if exists, or 'none — first cycle'}
+  Use previous cycle for trend comparison if available.",
+  model="{MODEL_MAP['qa-report-writer']}",
+  description="QA report cycle {N}"
+)
+```
+
+## 4. Write Completion Checkpoint
+
+```json
+{ "phase": "reporting_complete" }
+```
+
+## 5. Output Phase Transition
+
+Read and output `~/.claude/skills/qa/templates/stop-points/phase-transition.md`
+Substitute: {X} = 4, {Y} = 5
+
+</process>

package/skills/qa/workflows/test-phase.md

@@ -0,0 +1,86 @@
+# Workflow: Test Phase (Phase 3)
+
+<purpose>
+Execute test agents in parallel based on scope, save raw results to disk, and checkpoint state transitions.
+</purpose>
+
+<required_context>
+- PROFILE: .claude/qa-profile.json (apps, commands, scopes, agentRouting)
+- STATE: docs/qa-reports/cycle-state.json (cycle number, scope)
+- MODEL_MAP: resolved model per agent from ~/.claude/skills/qa/references/model-profiles.md
+</required_context>
+
+<process>
+
+## 1. Write State Checkpoint
+
+```json
+{ "phase": "testing", "cycle": N, "scope": "{scope}" }
+```
+
+Write to `docs/qa-reports/cycle-state.json` BEFORE spawning any agents.
+
+## 2. Create Raw Results Directory
+
+```bash
+mkdir -p docs/qa-reports/cycle-{N}-raw/
+```
+
+## 3. Determine Agents to Spawn
+
+Based on scope filter:
+
+| Scope | Agents |
+|-------|--------|
+| `full` | qa-test-executor, qa-security-auditor, qa-ux-optimizer |
+| `api` | qa-test-executor (backend apps only), qa-security-auditor |
+| `security` | qa-security-auditor |
+| `ux` | qa-ux-optimizer |
+| `{app-name}` | qa-test-executor (filtered to app) |
+
+## 4. Spawn Test Agents (parallel, up to 8)
+
+For each applicable agent, inline the following into the Task() prompt:
+- Agent instructions path: `.claude/agents/qa-{type}.md`
+- App list from PROFILE.apps (names, paths, frameworks, testIdPrefixes)
+- Test commands from PROFILE.commands
+- Cycle number and scope from STATE
+- Any scope-specific filtering instructions
+
+```
+Task(
+  subagent_type="general-purpose",
+  prompt="[QA-{AGENT-TYPE}] Read your instructions from .claude/agents/qa-{type}.md.
+  IMPORTANT: Read the Resources section FIRST — load all listed skill files before executing.
+
+  PROJECT CONTEXT (inlined from profile):
+  - Apps: {JSON array of relevant apps from PROFILE.apps}
+  - Test command: {PROFILE.commands.test}
+  - Cycle: {N}, Scope: {scope}
+  - Domain: {PROFILE.project.domain}
+  {scope-specific instructions}
+
+  Return results in structured table format.",
+  model="{MODEL_MAP[agent-type]}",
+  description="QA {agent-type} cycle {N}"
+)
+```
+
+## 5. Save Raw Results
+
+After each agent returns, save its output to:
+`docs/qa-reports/cycle-{N}-raw/{agent-type}.md`
+
+## 6. Write Completion Checkpoint
+
+After ALL agents return:
+```json
+{ "phase": "testing_complete" }
+```
+
+## 7. Output Phase Transition
+
+Read and output `~/.claude/skills/qa/templates/stop-points/phase-transition.md`
+Substitute: {X} = 3, {Y} = 4
+
+</process>

package/skills/qa/workflows/verify-phase.md

@@ -0,0 +1,65 @@
+# Workflow: Verify Phase (Phase 7b)
+
+<purpose>
+Run full verification after fix execution, then route to next test cycle.
+</purpose>
+
+<required_context>
+- PROFILE: .claude/qa-profile.json (commands.typeCheck, commands.build, commands.test)
+- STATE: docs/qa-reports/cycle-state.json (cycle number)
+- MODEL_MAP: resolved model for qa-verifier
+</required_context>
+
+<process>
+
+## 1. Full Build Verification
+
+Run sequentially:
+1. `{PROFILE.commands.typeCheck}`
+2. `{PROFILE.commands.build}`
+
+If either fails, log failure and continue to re-test (tests will catch the issues).
+
+## 2. Re-Run Full Test Suite
+
+Spawn qa-verifier agent:
+
+```
+Task(
+  subagent_type="general-purpose",
+  prompt="[QA-VERIFIER] Read .claude/agents/qa-verifier.md.
+  IMPORTANT: Read the Resources section FIRST.
+
+  Run FULL test suite (not just previously failed tests) to catch regressions.
+
+  PROJECT CONTEXT (inlined):
+  - Test command: {PROFILE.commands.test}
+  - Apps: {JSON summary of PROFILE.apps}
+  - Cycle: {N} (post-fix verification)
+
+  Return complete results in structured table format.",
+  model="{MODEL_MAP['qa-verifier']}",
+  description="QA full verify cycle {N}"
+)
+```
+
+## 3. Merge Results
+
+Merge verification results with cycle state.
+
+## 4. Increment Cycle Number
+
+```json
+{ "cycle": N+1 }
+```
+
+## 5. GOTO Phase 3
+
+Continue autonomously to next test cycle (Phase 3 in `/qa:full` context).
+
+## 6. Output Phase Transition
+
+Read and output `~/.claude/skills/qa/templates/stop-points/phase-transition.md`
+Substitute: {X} = 7b, {Y} = 3 (next cycle)
+
+</process>