claude-flow-novice 2.18.5 → 2.18.6

package/.claude/skills/cfn-test-framework/lib/execution/test-cache-reader.sh

@@ -0,0 +1,134 @@
+#!/usr/bin/env bash
+# test-cache-reader.sh
+# Worker-pattern test result reader: ONLY read cached results, NEVER run tests
+
+set -euo pipefail
+
+SWARM_ID="${1:-swarm-default}"
+AGENT_ID="${2:-worker-unknown}"
+REDIS_CHANNEL="swarm:${SWARM_ID}:tests"
+RESULTS_FILE="test-results.json"
+TIMEOUT="${TEST_CACHE_TIMEOUT:-300}"  # 5 minutes default
+
+echo "[WORKER:${AGENT_ID}] Waiting for coordinator test completion..."
+
+# 1. Wait for coordinator test completion signal
+ELAPSED=0
+COMPLETE=false
+
+while [ ${ELAPSED} -lt ${TIMEOUT} ]; do
+  # Check Redis for completion signal
+  if METADATA=$(redis-cli GET "${REDIS_CHANNEL}:metadata" 2>/dev/null); then
+    if [ -n "${METADATA}" ] && [ "${METADATA}" != "(nil)" ]; then
+      echo "[WORKER:${AGENT_ID}] Test completion signal received from Redis."
+      COMPLETE=true
+      break
+    fi
+  fi
+
+  # Fallback: Check file system signal
+  if [ -f ".test-results-ready" ] && [ -f "${RESULTS_FILE}" ]; then
+    echo "[WORKER:${AGENT_ID}] Test completion signal received from file system."
+    COMPLETE=true
+    break
+  fi
+
+  # Wait and retry
+  sleep 2
+  ELAPSED=$((ELAPSED + 2))
+
+  # Log progress every 30 seconds
+  if [ $((ELAPSED % 30)) -eq 0 ]; then
+    echo "[WORKER:${AGENT_ID}] Still waiting for test results... (${ELAPSED}s/${TIMEOUT}s)"
+  fi
+done
+
+if [ "${COMPLETE}" = false ]; then
+  echo "[WORKER:${AGENT_ID}] ERROR: Timeout waiting for test results after ${TIMEOUT}s" >&2
+  echo "[WORKER:${AGENT_ID}] Coordinator may not have executed tests yet." >&2
+  exit 1
+fi
+
+# 2. Read cached results (NEVER run tests)
+echo "[WORKER:${AGENT_ID}] Reading cached test results..."
+
+# Try Redis cache first (fastest)
+if CACHED_RESULTS=$(redis-cli GET "${REDIS_CHANNEL}:results" 2>/dev/null); then
+  if [ -n "${CACHED_RESULTS}" ] && [ "${CACHED_RESULTS}" != "(nil)" ]; then
+    echo "[WORKER:${AGENT_ID}] Using Redis-cached results."
+    echo "${CACHED_RESULTS}" > ".worker-${AGENT_ID}-results.json"
+    RESULTS_SOURCE="redis"
+  fi
+fi
+
+# Fallback to file cache
+if [ ! -f ".worker-${AGENT_ID}-results.json" ] && [ -f "${RESULTS_FILE}" ]; then
+  echo "[WORKER:${AGENT_ID}] Using file-cached results."
+  cp "${RESULTS_FILE}" ".worker-${AGENT_ID}-results.json"
+  RESULTS_SOURCE="file"
+fi
+
+# Verify results available
+if [ ! -f ".worker-${AGENT_ID}-results.json" ]; then
+  echo "[WORKER:${AGENT_ID}] ERROR: No cached results available" >&2
+  exit 1
+fi
+
+WORKER_RESULTS_FILE=".worker-${AGENT_ID}-results.json"
+
+# 3. Parse results for worker decision-making
+TESTS_PASSED=$(jq -r '.numPassedTests // 0' "${WORKER_RESULTS_FILE}" 2>/dev/null || echo "0")
+TESTS_FAILED=$(jq -r '.numFailedTests // 0' "${WORKER_RESULTS_FILE}" 2>/dev/null || echo "0")
+TESTS_TOTAL=$(jq -r '.numTotalTests // 0' "${WORKER_RESULTS_FILE}" 2>/dev/null || echo "0")
+COVERAGE_LINES=$(jq -r '.coverageMap.total.lines.pct // 0' "${WORKER_RESULTS_FILE}" 2>/dev/null || echo "0")
+COVERAGE_BRANCHES=$(jq -r '.coverageMap.total.branches.pct // 0' "${WORKER_RESULTS_FILE}" 2>/dev/null || echo "0")
+
+# 4. Output results summary
+echo "[WORKER:${AGENT_ID}] =========================================="
+echo "[WORKER:${AGENT_ID}] Test Results Summary (${RESULTS_SOURCE} cache)"
+echo "[WORKER:${AGENT_ID}] =========================================="
+echo "[WORKER:${AGENT_ID}] Tests:    ${TESTS_PASSED}/${TESTS_TOTAL} passed, ${TESTS_FAILED} failed"
+echo "[WORKER:${AGENT_ID}] Coverage: ${COVERAGE_LINES}% lines, ${COVERAGE_BRANCHES}% branches"
+echo "[WORKER:${AGENT_ID}] =========================================="
+
+# 5. Store parsed results for worker consumption
+cat > ".worker-${AGENT_ID}-summary.json" <<JSON
+{
+  "agent": "${AGENT_ID}",
+  "timestamp": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
+  "source": "${RESULTS_SOURCE}",
+  "tests": {
+    "passed": ${TESTS_PASSED},
+    "failed": ${TESTS_FAILED},
+    "total": ${TESTS_TOTAL}
+  },
+  "coverage": {
+    "lines": ${COVERAGE_LINES},
+    "branches": ${COVERAGE_BRANCHES}
+  }
+}
+JSON
+
+echo "[WORKER:${AGENT_ID}] Results cached to .worker-${AGENT_ID}-summary.json"
+
+# 6. Signal worker read completion to coordinator
+redis-cli LPUSH "${REDIS_CHANNEL}:workers:read" "$(cat <<JSON
+{
+  "agent": "${AGENT_ID}",
+  "timestamp": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
+  "source": "${RESULTS_SOURCE}"
+}
+JSON
+)" >/dev/null 2>&1 || true
+
+# 7. Return exit code based on test results
+if [ "${TESTS_FAILED}" -eq 0 ] && [ "${TESTS_TOTAL}" -gt 0 ]; then
+  echo "[WORKER:${AGENT_ID}] All tests passed. Worker can proceed."
+  exit 0
+elif [ "${TESTS_FAILED}" -gt 0 ]; then
+  echo "[WORKER:${AGENT_ID}] WARNING: ${TESTS_FAILED} tests failed. Worker should address failures."
+  exit 1
+else
+  echo "[WORKER:${AGENT_ID}] WARNING: No tests found. Worker should verify test execution."
+  exit 2
+fi

package/.claude/skills/cfn-test-framework/lib/execution/test-concurrent-conflicts.sh

@@ -0,0 +1,115 @@
+#!/bin/bash
+set -e
+
+RESULTS_FILE=".artifacts/analytics/test-conflict-results.json"
+mkdir -p .artifacts/analytics
+
+echo "═══════════════════════════════════════════════════════════"
+echo "🧪 Testing Concurrent Test Execution (20 runs)"
+echo "Coordinator Pattern: Single test run, multiple workers read cache"
+echo "═══════════════════════════════════════════════════════════"
+echo ""
+
+# Track conflicts
+CONFLICTS=0
+SUCCESSFUL_RUNS=0
+
+test_concurrent_execution() {
+  local run_id=$1
+
+  echo "──────────────────────────────────────"
+  echo "Test Cycle $run_id/20"
+  echo "──────────────────────────────────────"
+
+  # Step 1: Terminate existing tests (coordinator responsibility)
+  echo "[Run $run_id] Coordinator: Terminating existing tests..."
+  pkill -f vitest 2>/dev/null || true
+  pkill -f "npm test" 2>/dev/null || true
+  sleep 0.5
+
+  # Step 2: Coordinator creates test results (simulated)
+  echo "[Run $run_id] Coordinator: Running tests (simulated)..."
+  cat > test-results-$run_id.json << 'EOF'
+{
+  "numTotalTests": 100,
+  "numPassedTests": 95,
+  "numFailedTests": 5,
+  "numPendingTests": 0,
+  "testResults": [],
+  "success": true
+}
+EOF
+
+  # Step 3: Simulate 3 workers spawning and trying to access results
+  echo "[Run $run_id] Workers: Spawning 3 workers..."
+
+  for worker in {1..3}; do
+    (
+      # Worker waits for coordinator to finish
+      timeout 5 bash -c "
+        while [ ! -f test-results-$run_id.json ]; do
+          sleep 0.1
+        done
+        sleep 0.2 # Simulate processing time
+        echo '  Worker $worker: Read test results successfully'
+      " 2>/dev/null || echo "  Worker $worker: Timeout waiting for results"
+    ) &
+  done
+
+  # Wait for all workers
+  wait
+
+  # Step 4: Check for conflicts (multiple test processes)
+  VITEST_COUNT=$(pgrep -f vitest 2>/dev/null | wc -l || echo "0")
+  NPM_TEST_COUNT=$(pgrep -f "npm test" 2>/dev/null | wc -l || echo "0")
+  TOTAL_TEST_PROCESSES=$((VITEST_COUNT + NPM_TEST_COUNT))
+
+  if [ "$TOTAL_TEST_PROCESSES" -gt 0 ]; then
+    echo "❌ [Run $run_id] CONFLICT: $TOTAL_TEST_PROCESSES concurrent test processes detected"
+    CONFLICTS=$((CONFLICTS + 1))
+  else
+    echo "✅ [Run $run_id] No conflicts detected - coordinator pattern working"
+    SUCCESSFUL_RUNS=$((SUCCESSFUL_RUNS + 1))
+  fi
+
+  # Cleanup
+  pkill -f vitest 2>/dev/null || true
+  pkill -f "npm test" 2>/dev/null || true
+  rm -f test-results-$run_id.json
+  echo ""
+}
+
+# Run 20 test cycles
+for i in {1..20}; do
+  test_concurrent_execution $i
+done
+
+# Generate report
+CONFLICT_RATE=$(echo "scale=2; ($CONFLICTS / 20) * 100" | bc)
+
+cat > $RESULTS_FILE << REPORT
+{
+  "total_runs": 20,
+  "successful_runs": $SUCCESSFUL_RUNS,
+  "conflicts_detected": $CONFLICTS,
+  "conflict_rate": $CONFLICT_RATE,
+  "passed": $([ $CONFLICTS -eq 0 ] && echo "true" || echo "false"),
+  "threshold": "zero_conflicts",
+  "coordinator_pattern": "implemented",
+  "test_date": "$(date -I)"
+}
+REPORT
+
+echo "═══════════════════════════════════════════════════════════"
+echo "TEST RESULTS"
+echo "═══════════════════════════════════════════════════════════"
+cat $RESULTS_FILE
+echo ""
+
+if [ $CONFLICTS -eq 0 ]; then
+  echo "✅ SUCCESS: Zero concurrent test execution conflicts in 20 runs"
+  exit 0
+else
+  echo "❌ FAILED: $CONFLICTS conflicts detected in 20 runs (${CONFLICT_RATE}% conflict rate)"
+  exit 1
+fi

package/.claude/skills/cfn-test-framework/lib/execution/test-coordinator-pattern.sh

@@ -0,0 +1,109 @@
+#!/usr/bin/env bash
+# test-coordinator-pattern.sh
+# Coordinator-pattern test execution: Run tests ONCE, cache results for workers
+
+set -euo pipefail
+
+SWARM_ID="${1:-swarm-default}"
+REDIS_CHANNEL="swarm:${SWARM_ID}:tests"
+RESULTS_FILE="test-results.json"
+
+echo "[COORDINATOR] Starting test execution for swarm: ${SWARM_ID}"
+
+# 1. Terminate existing test runs to prevent conflicts
+echo "[COORDINATOR] Terminating existing test processes..."
+pkill -f vitest 2>/dev/null || true
+pkill -f "npm test" 2>/dev/null || true
+pkill -f "node.*test" 2>/dev/null || true
+sleep 1
+
+# Verify no test processes remain
+REMAINING=$(pgrep -f "vitest|npm test" | wc -l || echo "0")
+if [ "${REMAINING}" -gt 0 ]; then
+  echo "[COORDINATOR] WARNING: ${REMAINING} test processes still running. Force killing..."
+  pkill -9 -f vitest 2>/dev/null || true
+  pkill -9 -f "npm test" 2>/dev/null || true
+  sleep 1
+fi
+
+# 2. Execute tests once and cache results
+echo "[COORDINATOR] Running tests (single execution)..."
+START_TIME=$(date +%s)
+
+# Run tests with JSON reporter for parsing
+if npm test -- --run --reporter=json > "${RESULTS_FILE}" 2>&1; then
+  TEST_EXIT_CODE=0
+else
+  TEST_EXIT_CODE=$?
+fi
+
+END_TIME=$(date +%s)
+DURATION=$((END_TIME - START_TIME))
+
+echo "[COORDINATOR] Test execution complete in ${DURATION}s (exit code: ${TEST_EXIT_CODE})"
+
+# 3. Parse test results
+if [ -f "${RESULTS_FILE}" ]; then
+  TESTS_PASSED=$(jq -r '.numPassedTests // 0' "${RESULTS_FILE}" 2>/dev/null || echo "0")
+  TESTS_FAILED=$(jq -r '.numFailedTests // 0' "${RESULTS_FILE}" 2>/dev/null || echo "0")
+  TESTS_TOTAL=$(jq -r '.numTotalTests // 0' "${RESULTS_FILE}" 2>/dev/null || echo "0")
+  COVERAGE_LINES=$(jq -r '.coverageMap.total.lines.pct // 0' "${RESULTS_FILE}" 2>/dev/null || echo "0")
+  COVERAGE_BRANCHES=$(jq -r '.coverageMap.total.branches.pct // 0' "${RESULTS_FILE}" 2>/dev/null || echo "0")
+
+  echo "[COORDINATOR] Results: ${TESTS_PASSED}/${TESTS_TOTAL} passed, ${TESTS_FAILED} failed"
+  echo "[COORDINATOR] Coverage: ${COVERAGE_LINES}% lines, ${COVERAGE_BRANCHES}% branches"
+else
+  echo "[COORDINATOR] WARNING: ${RESULTS_FILE} not found. Using default values."
+  TESTS_PASSED=0
+  TESTS_FAILED=0
+  TESTS_TOTAL=0
+  COVERAGE_LINES=0
+  COVERAGE_BRANCHES=0
+fi
+
+# 4. Signal completion via Redis pub/sub
+COMPLETION_MESSAGE=$(cat <<JSON
+{
+  "event": "tests_complete",
+  "swarmId": "${SWARM_ID}",
+  "timestamp": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
+  "duration": ${DURATION},
+  "exitCode": ${TEST_EXIT_CODE},
+  "tests": {
+    "passed": ${TESTS_PASSED},
+    "failed": ${TESTS_FAILED},
+    "total": ${TESTS_TOTAL}
+  },
+  "coverage": {
+    "lines": ${COVERAGE_LINES},
+    "branches": ${COVERAGE_BRANCHES}
+  },
+  "resultsFile": "${RESULTS_FILE}"
+}
+JSON
+)
+
+echo "[COORDINATOR] Publishing completion signal to Redis..."
+redis-cli PUBLISH "${REDIS_CHANNEL}:complete" "${COMPLETION_MESSAGE}" >/dev/null 2>&1 || {
+  echo "[COORDINATOR] WARNING: Redis publish failed. Workers must rely on file cache."
+}
+
+# 5. Cache results in Redis (1-hour TTL)
+if [ -f "${RESULTS_FILE}" ]; then
+  echo "[COORDINATOR] Caching results in Redis (TTL: 3600s)..."
+  redis-cli SETEX "${REDIS_CHANNEL}:results" 3600 "$(cat "${RESULTS_FILE}")" >/dev/null 2>&1 || {
+    echo "[COORDINATOR] WARNING: Redis cache failed. Workers will use file cache."
+  }
+
+  # Store metadata separately for quick access
+  redis-cli SETEX "${REDIS_CHANNEL}:metadata" 3600 "${COMPLETION_MESSAGE}" >/dev/null 2>&1 || true
+fi
+
+# 6. Create worker-ready signal file (fallback for no Redis)
+touch ".test-results-ready"
+
+echo "[COORDINATOR] Test execution complete. Results cached for workers."
+echo "[COORDINATOR] Workers can now safely read results without running tests."
+
+# Return original test exit code
+exit ${TEST_EXIT_CODE}

package/.claude/skills/cfn-test-framework/lib/runner/SKILL.md

@@ -0,0 +1,288 @@
+# CFN Test Runner Skill
+
+**Version:** 1.0.0
+**Status:** Production
+**Purpose:** Unified test execution with benchmarking and regression detection
+
+---
+
+## Overview
+
+Provides comprehensive test execution across all CFN test suites:
+- **Hello World Tests** (Layer 0-7)
+- **CFN Loop E2E Tests** (9 integration tests)
+- **Benchmark Tracking** (SQLite storage)
+- **Regression Detection** (automated alerts)
+
+**Key Features:**
+- Single command test execution
+- Historical benchmark comparison
+- Performance regression alerts
+- Git-aware baseline tracking
+- Parallel test execution
+- Comprehensive reporting
+
+---
+
+## Architecture
+
+### Skill Components
+
+```
+.claude/skills/cfn-test-runner/
+├── SKILL.md                    # This file
+├── run-all-tests.sh            # Main test runner
+├── store-benchmarks.sh         # SQLite benchmark storage
+├── detect-regressions.sh       # Regression analysis
+├── init-benchmark-db.sh        # Database initialization
+└── generate-report.sh          # Test report generation
+```
+
+---
+
+## Usage
+
+### Run All Tests
+
+```bash
+./.claude/skills/cfn-test-runner/run-all-tests.sh \
+  --suite all \
+  --benchmark \
+  --detect-regressions
+```
+
+### Run Specific Suite
+
+```bash
+# Hello World only
+./.claude/skills/cfn-test-runner/run-all-tests.sh --suite hello-world
+
+# CFN E2E only
+./.claude/skills/cfn-test-runner/run-all-tests.sh --suite cfn-e2e
+
+# Both
+./.claude/skills/cfn-test-runner/run-all-tests.sh --suite all
+```
+
+### With Regression Detection
+
+```bash
+./.claude/skills/cfn-test-runner/run-all-tests.sh \
+  --suite all \
+  --benchmark \
+  --detect-regressions \
+  --threshold 0.10
+```
+
+---
+
+## Parameters
+
+### run-all-tests.sh
+
+| Parameter | Required | Description | Default |
+|-----------|----------|-------------|---------|
+| `--suite` | No | Test suite: `all`, `hello-world`, `cfn-e2e` | `all` |
+| `--benchmark` | No | Store results in SQLite | `false` |
+| `--detect-regressions` | No | Run regression analysis | `false` |
+| `--threshold` | No | Regression threshold (%) | `0.10` |
+| `--parallel` | No | Run tests in parallel | `false` |
+| `--output` | No | Output format: `text`, `json`, `html` | `text` |
+
+---
+
+## Test Suites
+
+### Hello World (Layer 0-7)
+
+**Purpose:** Validate agent spawning patterns and coordination
+
+| Layer | Test | Duration | Critical |
+|-------|------|----------|----------|
+| 0 | Tool Validation | 60s | ✅ |
+| 1-4 | [Future] | - | - |
+| 5 | Coordinator Spawning | 120s | ✅ |
+| 6 | Review Handoff | 180s | ✅ |
+| 7 | Error Retry | 150s | ✅ |
+
+**Location:** `tests/hello-world/`
+
+### CFN Loop E2E (9 Tests)
+
+**Purpose:** Validate full CFN Loop workflow
+
+| Test | Component | Duration | Critical |
+|------|-----------|----------|----------|
+| 1 | Coordinator → Orchestrator | 30s | ✅ |
+| 2 | Loop 3 → Gate Check | 60s | ✅ |
+| 3 | Gate Pass → Loop 2 | 30s | ⚠️ |
+| 4 | Loop 2 → Product Owner | 90s | ✅ |
+| 5 | Product Owner Decision | 60s | ✅ |
+| 6 | Iteration Cycle | 120s | ✅ |
+| 7 | Redis Key Structure | 10s | ✅ |
+| 8 | Error Recovery | 60s | ⚠️ |
+| 9 | Cleanup | 10s | ⚠️ |
+
+**Location:** `tests/cfn-v3/test-e2e-cfn-loop.sh`
+
+---
+
+## Benchmark Storage
+
+### SQLite Schema
+
+```sql
+test_suites          -- Suite definitions
+test_runs            -- Execution records
+test_cases           -- Individual tests
+test_results         -- Per-test outcomes
+performance_metrics  -- Performance data
+regression_alerts    -- Detected regressions
+```
+
+**Database:** `.artifacts/test-benchmarks.db`
+
+### Stored Metrics
+
+**Test-Level:**
+- Duration (ms)
+- Status (passed/failed/skipped)
+- Assertion count
+- Error messages
+
+**Suite-Level:**
+- Total duration (seconds)
+- Success rate (%)
+- Pass/fail/skip counts
+- Git commit/branch
+
+**Performance:**
+- Agent spawn time
+- Redis operation latency
+- File I/O duration
+- Memory usage
+
+---
+
+## Regression Detection
+
+### Automatic Alerts
+
+```bash
+# Critical: Test started failing
+alert_type: test_failure
+severity: critical
+message: "TEST 5: Product Owner Decision FAILED (was passing)"
+
+# Warning: Performance degraded
+alert_type: performance_regression
+severity: warning
+message: "Loop 3 spawn time increased 25% (baseline: 5s, current: 6.25s)"
+
+# Info: Success rate dropped
+alert_type: success_rate_drop
+severity: info
+message: "Suite success rate: 88% → 77% (threshold: 10%)"
+```
+
+### Threshold Configuration
+
+```bash
+# Default: 10% regression threshold
+--threshold 0.10
+
+# Strict: 5% threshold
+--threshold 0.05
+
+# Relaxed: 20% threshold
+--threshold 0.20
+```
+
+---
+
+## Output Formats
+
+### Text (Console)
+
+```
+==========================================
+CFN Test Suite Results
+==========================================
+
+Suite: Hello World
+  Layer 0: ✅ PASSED (58.3s)
+  Layer 5: ✅ PASSED (115.7s)
+  Layer 6: ✅ PASSED (172.4s)
+  Layer 7: ✅ PASSED (148.2s)
+
+Suite: CFN E2E
+  TEST 1: ✅ PASSED (28.1s)
+  TEST 2: ✅ PASSED (54.6s)
+  TEST 3: ⚠️  SKIPPED
+  TEST 4: ✅ PASSED (87.3s)
+  TEST 5: ✅ PASSED (59.2s)
+  TEST 6: ✅ PASSED (118.9s)
+  TEST 7: ✅ PASSED (9.4s)
+
+Total: 11 tests, 9 passed, 0 failed, 2 skipped
+Duration: 851.1s
+Success Rate: 81.8%
+
+Regressions Detected: 0
+==========================================
+```
+
+### JSON (API/CI)
+
+```json
+{
+  "timestamp": "2025-11-04T02:00:00Z",
+  "git_commit": "abc123",
+  "git_branch": "main",
+  "suites": {
+    "hello-world": {
+      "total": 4,
+      "passed": 4,
+      "failed": 0,
+      "skipped": 0,
+      "duration": 494.6,
+      "success_rate": 1.0
+    },
+    "cfn-e2e": {
+      "total": 9,
+      "passed": 7,
+      "failed": 0,
+      "skipped": 2,
+      "duration": 356.5,
+      "success_rate": 0.778
+    }
+  },
+  "regressions": [],
+  "baseline_comparison": {
+    "duration_change": "+2.3%",
+    "success_rate_change": "+0.0%"
+  }
+}
+```
+
+### HTML (Report)
+
+Generates interactive HTML report with:
+- Trend graphs (Chart.js)
+- Regression highlights
+- Test history
+- Performance metrics
+
+---
+
+## Baseline Management
+
+### Establish Baseline
+
+```bash
+# First run establishes baseline
+./.claude/skills/cfn-test-runner/run-all-tests.sh --suite all --benchmark
+
+# Mark as baseline
+sqlite3 .artifacts/test-benchmarks.db << EOF
+UPDATE test_runs SET environment = 'baseline' WHERE id = (SELECT MAX(id) FROM test_runs);

package/.claude/skills/cfn-test-framework/lib/runner/detect-regressions.sh

@@ -0,0 +1,58 @@
+#!/bin/bash
+# Detect test regressions
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PROJECT_ROOT="$(cd "$SCRIPT_DIR/../../.." && pwd)"
+
+# Load parameterized query library (SQL injection prevention)
+source "$PROJECT_ROOT/.claude/skills/bootstrap/sqlite-params.sh"
+
+DB_FILE="$PROJECT_ROOT/.artifacts/test-benchmarks.db"
+
+THRESHOLD=0.10
+
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    --threshold) THRESHOLD="$2"; shift 2 ;;
+    *) shift ;;
+  esac
+done
+
+# Get latest run
+LATEST_RUN=$(sqlite_select "$DB_FILE" "SELECT id FROM test_runs ORDER BY run_timestamp DESC LIMIT 1")
+
+if [ -z "$LATEST_RUN" ]; then
+  echo "No test runs found"
+  exit 0
+fi
+
+# Get baseline (average of last 10 runs excluding latest)
+BASELINE_SUCCESS_RATE=$(sqlite_select "$DB_FILE" "
+  SELECT AVG(success_rate) FROM (
+    SELECT success_rate FROM test_runs
+    WHERE id != ?1
+    ORDER BY run_timestamp DESC
+    LIMIT 10
+  )
+" "$LATEST_RUN")
+
+LATEST_SUCCESS_RATE=$(sqlite_select "$DB_FILE" "SELECT success_rate FROM test_runs WHERE id = ?1" "$LATEST_RUN")
+
+# Check for regression
+REGRESSION=$(awk "BEGIN {print ($BASELINE_SUCCESS_RATE - $LATEST_SUCCESS_RATE) > $THRESHOLD}")
+
+if [ "$REGRESSION" = "1" ]; then
+  DIFF=$(awk "BEGIN {printf \"%.1f\", ($BASELINE_SUCCESS_RATE - $LATEST_SUCCESS_RATE) * 100}")
+  MESSAGE="Success rate dropped ${DIFF}% (baseline: ${BASELINE_SUCCESS_RATE}, current: ${LATEST_SUCCESS_RATE})"
+
+  sqlite_exec "$DB_FILE" \
+    "INSERT INTO regression_alerts (run_id, alert_type, severity, message) VALUES (?1, ?2, ?3, ?4)" \
+    "$LATEST_RUN" "success_rate_drop" "warning" "$MESSAGE"
+
+  echo "⚠️  Regression detected: Success rate dropped ${DIFF}%"
+  exit 1
+fi
+
+echo "✅ No regressions detected"
+exit 0