qaa-agent 1.0.0 → 1.1.0

package/.claude/commands/qa-analyze.md

@@ -10,51 +10,12 @@ Analysis-only mode. Scans a repository, detects framework and stack, and produce
 - --dev-repo: explicit path to developer repository
 - --qa-repo: path to existing QA repository (produces gap analysis instead of blueprint)
 
-## What It Produces
-
-- SCAN_MANIFEST.md -- file tree, framework detection, testable surfaces
-- QA_ANALYSIS.md -- architecture overview, risk assessment, testing pyramid
-- TEST_INVENTORY.md -- prioritized test cases with IDs and explicit outcomes
-- QA_REPO_BLUEPRINT.md (if no QA repo) or GAP_ANALYSIS.md (if QA repo provided)
-
 ## Instructions
 
 1. Read `CLAUDE.md` -- all QA standards.
-2. Initialize pipeline context:
-   ```bash
-   node bin/qaa-tools.cjs init qa-start [user arguments]
-   ```
-3. Invoke scanner agent:
-
-Task(
-  prompt="
-    <objective>Scan repository and produce SCAN_MANIFEST.md</objective>
-    <execution_context>@agents/qaa-scanner.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
-
-4. Invoke analyzer agent:
-
-Task(
-  prompt="
-    <objective>Analyze repository and produce QA_ANALYSIS.md, TEST_INVENTORY.md, and blueprint or gap analysis</objective>
-    <execution_context>@agents/qaa-analyzer.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    - .qa-output/SCAN_MANIFEST.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
-
-5. Present results to user. No git operations. No test generation.
+2. Execute the workflow:
+
+Follow the workflow defined in `@workflows/qa-analyze.md` end-to-end.
+Preserve all workflow gates (scan verification, artifact checks).
 
 $ARGUMENTS

package/.claude/commands/qa-from-ticket.md

@@ -7,82 +7,17 @@ Generate test cases and executable test files from a ticket (Jira, Linear, GitHu
 /qa-from-ticket <ticket-source> [--dev-repo <path>]
 
 - ticket-source: one of:
-  - URL to Jira/Linear/GitHub issue (e.g., https://linear.app/team/PROJ-123)
+  - URL to GitHub/Jira/Linear issue
   - Plain text user story or acceptance criteria
   - File path to a .md or .txt with ticket details
 - --dev-repo: path to developer repository (default: current directory)
 
-## What It Produces
-
-- TEST_CASES_FROM_TICKET.md -- test cases derived from acceptance criteria
-- Test spec files (unit, API, E2E as appropriate)
-- Page Object Model files (for E2E tests if needed)
-- Fixture files (test data)
-- Traceability matrix: which test covers which acceptance criterion
-
 ## Instructions
 
 1. Read `CLAUDE.md` -- all QA standards.
+2. Execute the workflow:
 
-2. Parse the ticket source:
-
-   **If URL:** Fetch the ticket content using WebFetch or gh CLI (for GitHub issues).
-   ```bash
-   # GitHub Issues
-   gh issue view <number> --json title,body,labels,assignees
-
-   # For other URLs, use WebFetch to read the page
-   ```
-
-   **If plain text:** Use the text directly as the ticket content.
-
-   **If file path:** Read the file content.
-
-3. Extract from the ticket:
-   - **Title/Summary**: what the feature or bug is about
-   - **Acceptance Criteria**: the specific conditions that must be met (look for "AC:", "Given/When/Then", checkboxes, numbered criteria)
-   - **User Story**: "As a [role] I want [action] so that [benefit]"
-   - **Edge Cases**: any mentioned edge cases, error states, or constraints
-   - **Priority**: ticket priority if available (maps to test priority P0/P1/P2)
-
-4. Read the dev repo source code related to the ticket:
-   - Search for files matching keywords from the ticket title
-   - Read routes, controllers, services, models related to the feature
-   - Identify the actual implementation (or lack of it if the feature is not built yet)
-
-5. Generate test cases that map 1:1 to acceptance criteria:
-   - Each acceptance criterion becomes at least one test case
-   - Add negative tests for each criterion (what should NOT happen)
-   - Add edge case tests if mentioned in the ticket
-   - Every test case follows CLAUDE.md rules: unique ID, concrete inputs, explicit expected outcome
-
-6. Write TEST_CASES_FROM_TICKET.md with:
-   ```markdown
-   # Test Cases from Ticket: [TICKET-ID] [Title]
-
-   ## Source
-   - Ticket: [URL or "manual input"]
-   - Date: [YYYY-MM-DD]
-
-   ## Acceptance Criteria Mapping
-
-   | AC # | Criterion | Test ID(s) | Status |
-   |------|-----------|------------|--------|
-   | AC-1 | [criterion text] | UT-XXX-001, E2E-XXX-001 | Covered |
-   | AC-2 | [criterion text] | API-XXX-001 | Covered |
-
-   ## Test Cases
-   [test cases following CLAUDE.md format]
-   ```
-
-7. Generate executable test files using the qa-template-engine skill.
-
-8. Validate generated tests using the qa-self-validator skill.
-
-9. Present summary:
-   - How many acceptance criteria found
-   - How many test cases generated (by pyramid level)
-   - Traceability: every AC is covered
-   - Any ACs that could not be tested (and why)
+Follow the workflow defined in `@workflows/qa-from-ticket.md` end-to-end.
+Preserve all workflow gates (ticket parsing, traceability, validation).
 
 $ARGUMENTS

package/.claude/commands/qa-gap.md

@@ -9,46 +9,12 @@ Compare a developer repository against its QA repository to identify coverage ga
 - --dev-repo: path to the developer repository (required)
 - --qa-repo: path to the existing QA repository (required)
 
-## What It Produces
-
-- SCAN_MANIFEST.md -- scan of both repositories
-- GAP_ANALYSIS.md -- coverage map, missing tests with IDs, broken tests, quality assessment
-
 ## Instructions
 
 1. Read `CLAUDE.md` -- testing pyramid, test spec rules, quality gates.
-2. Invoke scanner agent to scan both repositories:
-
-Task(
-  prompt="
-    <objective>Scan both developer and QA repositories and produce SCAN_MANIFEST.md</objective>
-    <execution_context>@agents/qaa-scanner.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
-
-3. Invoke analyzer agent in gap mode:
-
-Task(
-  prompt="
-    <objective>Produce GAP_ANALYSIS.md comparing dev repo against QA repo</objective>
-    <execution_context>@agents/qaa-analyzer.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    - .qa-output/SCAN_MANIFEST.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    mode: gap
-    </parameters>
-  "
-)
-
-4. Present results. No test generation. No git operations.
+2. Execute the workflow:
+
+Follow the workflow defined in `@workflows/qa-gap.md` end-to-end.
+Preserve all workflow gates (both-repo scanning, gap metrics).
 
 $ARGUMENTS

package/.claude/commands/qa-start.md

@@ -14,20 +14,9 @@ Run the complete QA automation pipeline. Analyzes a repository, generates a stan
 ## Instructions
 
 1. Read `CLAUDE.md` -- all QA standards that govern the pipeline.
-2. Read `agents/qa-pipeline-orchestrator.md` -- the pipeline controller.
-3. Invoke the orchestrator:
-
-Task(
-  prompt="
-    <objective>Run complete QA automation pipeline</objective>
-    <execution_context>@agents/qa-pipeline-orchestrator.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
+2. Execute the workflow:
+
+Follow the workflow defined in `@workflows/qa-start.md` end-to-end.
+Preserve all workflow gates (stage transitions, checkpoints, error handling, state updates, commits).
 
 $ARGUMENTS

package/.claude/commands/qa-testid.md

@@ -4,51 +4,16 @@ Scan frontend source code, audit missing data-testid attributes, and inject them
 
 ## Usage
 
-/qa-testid <path-to-frontend-source>
+/qa-testid [<source-directory>]
 
-- path-to-frontend-source: directory containing React/Vue/Angular/HTML components
-
-## What It Produces
-
-- TESTID_AUDIT_REPORT.md -- coverage score, missing elements, proposed values by priority
-- Modified source files with data-testid attributes injected
+- source-directory: path to frontend source (auto-detects if omitted)
 
 ## Instructions
 
-1. Read `CLAUDE.md` -- data-testid Convention section for naming rules.
-2. Initialize context:
-   ```bash
-   node bin/qaa-tools.cjs init qa-start --dev-repo [user path]
-   ```
-3. Invoke scanner to identify component files:
-
-Task(
-  prompt="
-    <objective>Scan repository to identify frontend component files</objective>
-    <execution_context>@agents/qaa-scanner.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
-
-4. Invoke testid-injector agent:
-
-Task(
-  prompt="
-    <objective>Audit missing data-testid attributes and inject following naming convention</objective>
-    <execution_context>@agents/qaa-testid-injector.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    - .qa-output/SCAN_MANIFEST.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
+1. Read `CLAUDE.md` -- data-testid Convention section.
+2. Execute the workflow:
+
+Follow the workflow defined in `@workflows/qa-testid.md` end-to-end.
+Preserve all workflow gates (framework detection, user approval, branch creation).
 
 $ARGUMENTS

package/.claude/commands/qa-validate.md

@@ -4,51 +4,17 @@ Validate existing test files against CLAUDE.md standards. Runs 4-layer validatio
 
 ## Usage
 
-/qa-validate <path-to-tests> [--framework <name>]
+/qa-validate [<test-directory>] [--classify]
 
-- path-to-tests: directory or specific test files to validate
-- --framework: override framework auto-detection (playwright, cypress, jest, etc.)
-
-## What It Produces
-
-- VALIDATION_REPORT.md -- pass/fail per file per validation layer, confidence level
-- FAILURE_CLASSIFICATION_REPORT.md -- if failures found, classifies as APP BUG / TEST ERROR / ENV ISSUE / INCONCLUSIVE
+- test-directory: path to test files (auto-detects if omitted)
+- --classify: also run bug-detective to classify failures
 
 ## Instructions
 
 1. Read `CLAUDE.md` -- quality gates, locator tiers, assertion rules.
-2. Invoke validator agent:
-
-Task(
-  prompt="
-    <objective>Validate test files with 4-layer validation and produce VALIDATION_REPORT.md</objective>
-    <execution_context>@agents/qaa-validator.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    mode: validation
-    </parameters>
-  "
-)
-
-3. If failures detected, invoke bug-detective agent:
-
-Task(
-  prompt="
-    <objective>Classify test failures and auto-fix TEST CODE ERRORS</objective>
-    <execution_context>@agents/qaa-bug-detective.md</execution_context>
-    <files_to_read>
-    - CLAUDE.md
-    - .qa-output/VALIDATION_REPORT.md
-    </files_to_read>
-    <parameters>
-    user_input: $ARGUMENTS
-    </parameters>
-  "
-)
-
-4. Present results. No git operations.
+2. Execute the workflow:
+
+Follow the workflow defined in `@workflows/qa-validate.md` end-to-end.
+Preserve all workflow gates (fix loops, classification).
 
 $ARGUMENTS

package/bin/install.cjs

@@ -106,6 +106,12 @@ async function main() {
   const skillCount = copyDir(skillsSrc, skillsDest);
   ok(`Installed ${skillCount} skill files (6 skills)`);
 
+  // Install workflows
+  const workflowsSrc = path.join(ROOT, 'workflows');
+  const workflowsDest = path.join(qaaDir, 'workflows');
+  const wfCount = copyDir(workflowsSrc, workflowsDest);
+  ok(`Installed ${wfCount} workflows`);
+
   // Install agents
   const agentsSrc = path.join(ROOT, 'agents');
   const agentsDest = path.join(qaaDir, 'agents');

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "qaa-agent",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "QA Automation Agent for Claude Code — multi-agent pipeline that analyzes repos, generates tests, validates, and creates PRs",
   "bin": {
     "qaa-agent": "./bin/install.cjs"
@@ -24,6 +24,7 @@
   "files": [
     "bin/",
     "agents/",
+    "workflows/",
     "templates/",
     ".claude/commands/",
     ".claude/skills/",

package/workflows/qa-analyze.md

@@ -0,0 +1,296 @@
+<purpose>
+Analysis-only workflow. Scans a developer repository (and optionally a QA repository) to produce comprehensive analysis artifacts -- SCAN_MANIFEST.md, QA_ANALYSIS.md, TEST_INVENTORY.md, and optionally QA_REPO_BLUEPRINT.md or GAP_ANALYSIS.md. No test generation, no git operations, no PR. Use this workflow to understand a codebase's testability before committing to full test generation.
+</purpose>
+
+<required_reading>
+- `CLAUDE.md` -- QA automation standards, pipeline stages, module boundaries, verification commands
+- `agents/qaa-scanner.md` -- Scanner agent definition (produces SCAN_MANIFEST.md)
+- `agents/qaa-analyzer.md` -- Analyzer agent definition (produces QA_ANALYSIS.md, TEST_INVENTORY.md, QA_REPO_BLUEPRINT.md or GAP_ANALYSIS.md)
+- `templates/scan-manifest.md` -- SCAN_MANIFEST.md format contract
+- `templates/qa-analysis.md` -- QA_ANALYSIS.md format contract
+- `templates/test-inventory.md` -- TEST_INVENTORY.md format contract
+- `templates/qa-repo-blueprint.md` -- QA_REPO_BLUEPRINT.md format contract (Option 1 only)
+- `templates/gap-analysis.md` -- GAP_ANALYSIS.md format contract (Option 2/3 only)
+</required_reading>
+
+<process>
+
+<step name="parse_arguments">
+## Step 1: Parse Arguments
+
+Parse `$ARGUMENTS` for repository paths.
+
+**Supported arguments:**
+- `--dev-repo <path>` -- Path to the developer repository to analyze
+- `--qa-repo <path>` -- Path to an existing QA repository (optional)
+- No arguments -- defaults to the current working directory as the dev repo
+
+**Parsing logic:**
+
+```bash
+DEV_REPO=""
+QA_REPO=""
+
+# Parse --dev-repo and --qa-repo from $ARGUMENTS
+# If --dev-repo is provided, use it
+# If --qa-repo is provided, use it
+# If neither is provided, DEV_REPO defaults to current working directory
+```
+
+**Validation:**
+- If `--dev-repo` is provided, verify the path exists and is a directory. If not, print error: `"Error: Dev repo path does not exist: {path}"` and STOP.
+- If `--qa-repo` is provided, verify the path exists and is a directory. If not, print error: `"Error: QA repo path does not exist: {path}"` and STOP.
+- If no arguments provided, set `DEV_REPO` to the current working directory.
+</step>
+
+<step name="determine_mode">
+## Step 2: Determine Analysis Mode
+
+Set the workflow option based on whether a QA repo was provided.
+
+**Mode selection:**
+
+| QA Repo Provided | Mode | Description |
+|-------------------|------|-------------|
+| No | `full` (Option 1) | Dev-only analysis. Produces QA_ANALYSIS.md + TEST_INVENTORY.md + QA_REPO_BLUEPRINT.md |
+| Yes | `gap` (Option 2/3) | Gap analysis. Produces QA_ANALYSIS.md + TEST_INVENTORY.md + GAP_ANALYSIS.md |
+
+```
+MODE="full"
+if QA_REPO is not empty:
+  MODE="gap"
+```
+
+**Set output directory:**
+
+```bash
+OUTPUT_DIR=".qa-output"
+mkdir -p "${OUTPUT_DIR}"
+```
+
+**Print analysis banner:**
+
+```
+=== QA Analysis Workflow ===
+Mode: {MODE} ({description})
+Dev Repo: {DEV_REPO}
+QA Repo: {QA_REPO or 'N/A'}
+Output: {OUTPUT_DIR}
+============================
+```
+</step>
+
+<step name="run_scanner">
+## Step 3: Spawn Scanner Agent
+
+Spawn the scanner agent to produce SCAN_MANIFEST.md.
+
+**For full mode (Option 1):**
+
+```
+Task(
+  prompt="
+    <objective>Scan repository and produce SCAN_MANIFEST.md</objective>
+    <execution_context>@agents/qaa-scanner.md</execution_context>
+    <files_to_read>
+    - CLAUDE.md
+    </files_to_read>
+    <parameters>
+    dev_repo_path: {DEV_REPO}
+    qa_repo_path: null
+    output_path: {OUTPUT_DIR}/SCAN_MANIFEST.md
+    </parameters>
+  "
+)
+```
+
+**For gap mode (Option 2/3):**
+
+```
+Task(
+  prompt="
+    <objective>Scan both developer and QA repositories and produce SCAN_MANIFEST.md</objective>
+    <execution_context>@agents/qaa-scanner.md</execution_context>
+    <files_to_read>
+    - CLAUDE.md
+    </files_to_read>
+    <parameters>
+    dev_repo_path: {DEV_REPO}
+    qa_repo_path: {QA_REPO}
+    output_path: {OUTPUT_DIR}/SCAN_MANIFEST.md
+    </parameters>
+  "
+)
+```
+
+**Handle scanner return:**
+
+- If scanner returns `decision: STOP` -- print the stop reason and STOP the workflow. The repository has no testable surfaces or an uncertain framework.
+- If scanner returns `decision: PROCEED` -- continue to next step.
+- Extract `has_frontend` and `detection_confidence` from scanner return for use in the summary.
+
+**Verify SCAN_MANIFEST.md exists:**
+
+```bash
+[ -f "${OUTPUT_DIR}/SCAN_MANIFEST.md" ] && echo "FOUND" || echo "MISSING"
+```
+
+If missing, print error: `"Error: Scanner did not produce SCAN_MANIFEST.md. Check scanner output for details."` and STOP.
+</step>
+
+<step name="run_analyzer">
+## Step 4: Spawn Analyzer Agent
+
+Spawn the analyzer agent to produce analysis artifacts.
+
+**For full mode (Option 1):**
+
+```
+Task(
+  prompt="
+    <objective>Analyze scanned repository and produce QA_ANALYSIS.md, TEST_INVENTORY.md, and QA_REPO_BLUEPRINT.md</objective>
+    <execution_context>@agents/qaa-analyzer.md</execution_context>
+    <files_to_read>
+    - {OUTPUT_DIR}/SCAN_MANIFEST.md
+    - CLAUDE.md
+    </files_to_read>
+    <parameters>
+    workflow_option: 1
+    qa_analysis_path: {OUTPUT_DIR}/QA_ANALYSIS.md
+    test_inventory_path: {OUTPUT_DIR}/TEST_INVENTORY.md
+    blueprint_path: {OUTPUT_DIR}/QA_REPO_BLUEPRINT.md
+    </parameters>
+  "
+)
+```
+
+**For gap mode (Option 2/3):**
+
+```
+Task(
+  prompt="
+    <objective>Analyze scanned repositories and produce QA_ANALYSIS.md, TEST_INVENTORY.md, and GAP_ANALYSIS.md</objective>
+    <execution_context>@agents/qaa-analyzer.md</execution_context>
+    <files_to_read>
+    - {OUTPUT_DIR}/SCAN_MANIFEST.md
+    - CLAUDE.md
+    </files_to_read>
+    <parameters>
+    workflow_option: 2
+    qa_analysis_path: {OUTPUT_DIR}/QA_ANALYSIS.md
+    test_inventory_path: {OUTPUT_DIR}/TEST_INVENTORY.md
+    gap_analysis_path: {OUTPUT_DIR}/GAP_ANALYSIS.md
+    </parameters>
+  "
+)
+```
+
+**Handle analyzer return:**
+
+- Extract `total_test_count`, `pyramid_breakdown`, and `risk_count` from the analyzer's return values.
+- Verify all expected artifacts exist on disk.
+
+**Verify artifacts exist:**
+
+```bash
+[ -f "${OUTPUT_DIR}/QA_ANALYSIS.md" ] && echo "FOUND: QA_ANALYSIS.md" || echo "MISSING: QA_ANALYSIS.md"
+[ -f "${OUTPUT_DIR}/TEST_INVENTORY.md" ] && echo "FOUND: TEST_INVENTORY.md" || echo "MISSING: TEST_INVENTORY.md"
+
+# For Option 1 only:
+[ -f "${OUTPUT_DIR}/QA_REPO_BLUEPRINT.md" ] && echo "FOUND: QA_REPO_BLUEPRINT.md" || echo "MISSING: QA_REPO_BLUEPRINT.md"
+
+# For Option 2/3 only:
+[ -f "${OUTPUT_DIR}/GAP_ANALYSIS.md" ] && echo "FOUND: GAP_ANALYSIS.md" || echo "MISSING: GAP_ANALYSIS.md"
+```
+
+If any required artifact is missing, print error with the specific missing file and STOP.
+</step>
+
+<step name="print_summary">
+## Step 5: Print Analysis Summary
+
+Print a human-readable summary of the analysis results. No git operations are performed.
+
+**Read QA_ANALYSIS.md to extract summary data:**
+
+- Count risks by severity: HIGH, MEDIUM, LOW
+- Count test cases by pyramid tier from TEST_INVENTORY.md
+- Extract Top 10 unit test targets
+- Extract architecture overview (framework, language, runtime)
+
+**Print summary:**
+
+```
+=== Analysis Complete ===
+
+Architecture:
+  Framework: {framework}
+  Language: {language}
+  Runtime: {runtime}
+  Frontend: {has_frontend}
+  Detection Confidence: {detection_confidence}
+
+Risk Assessment:
+  HIGH: {high_count}
+  MEDIUM: {medium_count}
+  LOW: {low_count}
+
+Test Case Inventory:
+  Unit Tests:        {unit_count} ({unit_percent}%)
+  Integration Tests: {integration_count} ({integration_percent}%)
+  API Tests:         {api_count} ({api_percent}%)
+  E2E Tests:         {e2e_count} ({e2e_percent}%)
+  --------------------------
+  Total:             {total_count}
+
+Priority Distribution:
+  P0 (blocks release): {p0_count}
+  P1 (should fix):     {p1_count}
+  P2 (nice to have):   {p2_count}
+
+Artifacts Produced:
+  - {OUTPUT_DIR}/SCAN_MANIFEST.md
+  - {OUTPUT_DIR}/QA_ANALYSIS.md
+  - {OUTPUT_DIR}/TEST_INVENTORY.md
+  - {OUTPUT_DIR}/QA_REPO_BLUEPRINT.md  (Option 1 only)
+  - {OUTPUT_DIR}/GAP_ANALYSIS.md       (Option 2/3 only)
+
+No git operations performed. No PR created.
+Run the full pipeline with /qa-start to generate tests.
+===========================
+```
+</step>
+
+</process>
+
+<output>
+This workflow produces analysis artifacts only. No test files are generated. No git operations are performed.
+
+**Artifacts produced:**
+
+| Artifact | When Produced | Description |
+|----------|---------------|-------------|
+| SCAN_MANIFEST.md | Always | Repository scan with file tree, framework detection, testable surfaces |
+| QA_ANALYSIS.md | Always | Architecture overview, risks, top 10 targets, API targets, testing pyramid |
+| TEST_INVENTORY.md | Always | Complete test case inventory with IDs, inputs, expected outcomes |
+| QA_REPO_BLUEPRINT.md | Option 1 (no QA repo) | Repository structure, configs, CI/CD strategy for new QA repo |
+| GAP_ANALYSIS.md | Option 2/3 (QA repo exists) | Coverage gaps, missing tests, broken tests, quality assessment |
+
+**No side effects:**
+- No git branches created
+- No git commits made
+- No PRs created
+- No test files generated
+- No source files modified
+</output>
+
+<error_handling>
+| Error | Cause | Action |
+|-------|-------|--------|
+| Dev repo path does not exist | Invalid --dev-repo argument | Print error with path, STOP |
+| QA repo path does not exist | Invalid --qa-repo argument | Print error with path, STOP |
+| Scanner returns STOP | No testable surfaces or uncertain framework | Print scanner's reason, STOP |
+| SCAN_MANIFEST.md missing after scanner | Scanner failed silently | Print error, STOP |
+| Analyzer artifacts missing | Analyzer failed silently | Print error with specific missing file, STOP |
+| Framework detection LOW confidence | Ambiguous tech stack | Scanner checkpoints for user confirmation |
+</error_handling>