specweave 1.0.550 → 1.0.552

package/plugins/specweave/hooks/_archive/stop-auto-v4-legacy.sh

@@ -1,1319 +0,0 @@
-#!/bin/bash
-# stop-auto.sh - Auto Mode Stop Hook (v4.0 - Quality Gates & Auto-Close)
-#
-# PHILOSOPHY: Smart completion detection with quality gates:
-#   1. Is this a SpecWeave project with auto enabled?
-#   2. Are there active increments?
-#   3. For each active increment:
-#      a. Are all tasks marked complete in tasks.md?
-#      b. If TDD mode: Do tests pass?
-#      c. Are quality gates satisfied (ACs, coverage)?
-#      d. If YES to all → AUTO-CLOSE (triggers external sync!)
-#      e. If NO → Block with specific reason
-#
-# v4.0 CRITICAL FIX:
-# - Actually validates completion, not just metadata status
-# - Runs tests when TDD/requireTests mode is on
-# - Auto-closes increments that pass validation → triggers GitHub sync!
-# - Stop hook now has TEETH, not just a message
-#
-# Configuration is read from .specweave/config.json:
-#   - auto.enabled: Master switch (default: true)
-#   - auto.requireTests: Run tests before auto-close (default: false)
-#   - testing.defaultTestMode: "tdd" enables strict test enforcement
-#   - auto.skipQualityGates: Skip validation (DANGEROUS, default: false)
-#
-# This implements SpecWeave's autonomous execution pattern with real validation.
-
-set +e  # Don't exit on errors
-
-# Capture start time for duration tracking (macOS doesn't support %N, fallback to seconds only)
-if [[ "$OSTYPE" == "darwin"* ]]; then
-    _START_TIME_MS=$(($(date +%s) * 1000))
-else
-    _START_TIME_MS=$(($(date +%s) * 1000 + $(date +%N 2>/dev/null | cut -c1-3 || echo "0")))
-fi
-
-# Read stdin (Claude Code passes context here)
-cat > /dev/null  # Consume stdin (unused by this hook)
-PROJECT_ROOT="${PROJECT_ROOT:-$(pwd)}"
-
-# ============================================================================
-# SOURCE STRUCTURED DECISION LOGGING
-# ============================================================================
-SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
-if [ -f "$SCRIPT_DIR/log-decision.sh" ]; then
-    source "$SCRIPT_DIR/log-decision.sh"
-fi
-
-# Helper to calculate duration in ms (macOS-compatible)
-_get_duration_ms() {
-    local end_time_ms
-    if [[ "$OSTYPE" == "darwin"* ]]; then
-        end_time_ms=$(($(date +%s) * 1000))
-    else
-        end_time_ms=$(($(date +%s) * 1000 + $(date +%N 2>/dev/null | cut -c1-3 || echo "0")))
-    fi
-    echo $((end_time_ms - _START_TIME_MS))
-}
-
-# ============================================================================
-# LOGGING
-# ============================================================================
-LOGS_DIR="$PROJECT_ROOT/.specweave/logs"
-mkdir -p "$LOGS_DIR" 2>/dev/null
-LOG_FILE="$LOGS_DIR/stop-auto.log"
-
-log() {
-    echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] $1" >> "$LOG_FILE" 2>/dev/null
-    if [ "${SPECWEAVE_DEBUG_HOOKS:-0}" = "1" ]; then
-        echo -e "\033[36m[stop-auto]\033[0m $1" >&2
-    fi
-}
-
-# ============================================================================
-# SILENT APPROVE - Normal sessions get NO output
-# ============================================================================
-silent_approve() {
-    local reason="$1"
-    local reason_code="${2:-session_inactive}"
-    local context_json="${3:-"{}"}"
-
-    log "APPROVE: $reason"
-
-    # Log structured decision if log_decision function is available
-    if type log_decision &>/dev/null; then
-        log_decision "stop-auto" "approve" "$reason_code" "$reason" "$context_json" "$(_get_duration_ms)"
-    fi
-
-    echo '{"decision":"approve"}'
-    exit 0
-}
-
-# ============================================================================
-# LOUD APPROVE - Session completions need user notification
-# Used when work completes, increments close, or session ends successfully
-# ============================================================================
-loud_approve() {
-    local reason="$1"
-    local reason_code="${2:-session_complete}"
-    local context_json="${3:-"{}"}"
-    local message="$4"
-
-    log "APPROVE (loud): $reason"
-
-    # Log structured decision if log_decision function is available
-    if type log_decision &>/dev/null; then
-        log_decision "stop-auto" "approve" "$reason_code" "$reason" "$context_json" "$(_get_duration_ms)"
-    fi
-
-    jq -n \
-        --arg decision "approve" \
-        --arg reason "$reason" \
-        --arg msg "$message" \
-        '{decision: $decision, reason: $reason, systemMessage: $msg}'
-    exit 0
-}
-
-# ============================================================================
-# QUICK EXITS - Not a SpecWeave project
-# ============================================================================
-
-SPECWEAVE_DIR="$PROJECT_ROOT/.specweave"
-INCREMENTS_DIR="$SPECWEAVE_DIR/increments"
-CONFIG_FILE="$SPECWEAVE_DIR/config.json"
-
-# Not a SpecWeave project - silent approve
-[ ! -d "$SPECWEAVE_DIR" ] && silent_approve "Not a SpecWeave project" "not_specweave_project" "{}"
-[ ! -d "$INCREMENTS_DIR" ] && silent_approve "No increments directory" "no_increments_dir" "{}"
-
-# ============================================================================
-# STATE DIRECTORY (MUST be defined BEFORE AUTO_SESSION_FILE check)
-# ============================================================================
-
-STATE_DIR="$SPECWEAVE_DIR/state"
-
-# ============================================================================
-# READ PROJECT CONFIG
-# ============================================================================
-
-AUTO_ENABLED="true"
-REQUIRE_TESTS="false"
-REQUIRE_VALIDATION="true"
-REQUIRE_JUDGE_LLM="false"
-REQUIRE_LLM_EVAL="false"    # NEW: LLM-based completion evaluation
-MAX_TURNS="20"           # HARD STOP: Total turns in session (NEVER resets during session)
-MAX_RETRIES="20"         # Stuck detection: Retries on same work (resets when work changes)
-TDD_MODE="false"
-SKIP_QUALITY_GATES="false"
-TEST_COMMAND=""
-MAX_SESSION_AGE="7200"   # Stale session timeout (default 2 hours)
-
-if [ -f "$CONFIG_FILE" ]; then
-    AUTO_ENABLED=$(jq -r '.auto.enabled // true' "$CONFIG_FILE" 2>/dev/null || echo "true")
-    REQUIRE_TESTS=$(jq -r '.auto.requireTests // false' "$CONFIG_FILE" 2>/dev/null || echo "false")
-    REQUIRE_VALIDATION=$(jq -r '.auto.requireValidation // true' "$CONFIG_FILE" 2>/dev/null || echo "true")
-    REQUIRE_JUDGE_LLM=$(jq -r '.auto.requireJudgeLLM // false' "$CONFIG_FILE" 2>/dev/null || echo "false")
-    REQUIRE_LLM_EVAL=$(jq -r '.auto.requireLLMEval // false' "$CONFIG_FILE" 2>/dev/null || echo "false")
-    # maxTurns = HARD STOP (total turns in session, NEVER resets) - default 20
-    MAX_TURNS=$(jq -r '.auto.maxTurns // 20' "$CONFIG_FILE" 2>/dev/null || echo "20")
-    # maxRetries = stuck detection (resets when increments change) - default 20
-    MAX_RETRIES=$(jq -r '.auto.maxRetries // 20' "$CONFIG_FILE" 2>/dev/null || echo "20")
-    TEST_COMMAND=$(jq -r '.auto.testCommand // ""' "$CONFIG_FILE" 2>/dev/null || echo "")
-    SKIP_QUALITY_GATES=$(jq -r '.auto.skipQualityGates // false' "$CONFIG_FILE" 2>/dev/null || echo "false")
-    TDD_MODE_CONFIG=$(jq -r '.testing.defaultTestMode // "standard"' "$CONFIG_FILE" 2>/dev/null || echo "standard")
-    [ "$TDD_MODE_CONFIG" = "tdd" ] || [ "$TDD_MODE_CONFIG" = "TDD" ] && TDD_MODE="true"
-    # maxSessionAge = stale session timeout in seconds (default 2 hours = 7200s)
-    MAX_SESSION_AGE=$(jq -r '.auto.maxSessionAge // 7200' "$CONFIG_FILE" 2>/dev/null || echo "7200")
-fi
-
-# Auto mode disabled in config - silent approve
-[ "$AUTO_ENABLED" != "true" ] && silent_approve "Auto mode disabled in config" "auto_disabled" "{}"
-
-# ============================================================================
-# CHECK FOR AUTO MODE SESSION - Only block if explicitly activated
-# Auto mode is SESSION-SCOPED: if Claude Code session ends, auto mode ends.
-# ============================================================================
-
-AUTO_SESSION_FILE="$STATE_DIR/auto-mode.json"
-
-# If no auto-mode.json exists, auto mode was never started this session
-if [ ! -f "$AUTO_SESSION_FILE" ]; then
-    silent_approve "Auto mode not activated (no session file)" "session_inactive" '{"sessionActive":false}'
-fi
-
-# STALENESS CHECK: If session file is older than 30 minutes, it's stale
-# (A real auto mode session would have activity within 30 minutes)
-FILE_MTIME=$(stat -f%m "$AUTO_SESSION_FILE" 2>/dev/null || stat -c%Y "$AUTO_SESSION_FILE" 2>/dev/null || echo "0")
-CURRENT_TIME=$(date +%s)
-SESSION_AGE=$((CURRENT_TIME - FILE_MTIME))
-# MAX_SESSION_AGE loaded from config above (default: 7200s = 2 hours)
-
-if [ "$SESSION_AGE" -gt "$MAX_SESSION_AGE" ]; then
-    log "STALE SESSION DETECTED: Session file is ${SESSION_AGE}s old (max: ${MAX_SESSION_AGE}s)"
-    # Clean up stale session and ALL state files
-    rm -f "$AUTO_SESSION_FILE" 2>/dev/null
-    rm -f "$STATE_DIR/.stop-auto-dedup" 2>/dev/null
-    rm -f "$STATE_DIR/.stop-auto-retry" 2>/dev/null
-    rm -f "$STATE_DIR/.stop-auto-turns" 2>/dev/null  # Also clear turn counter
-    silent_approve "Stale auto-mode session cleared (inactive for ${SESSION_AGE}s)" "session_stale" "$(jq -n --argjson age "$SESSION_AGE" --argjson maxAge "$MAX_SESSION_AGE" '{sessionAge:$age,maxSessionAge:$maxAge}')"
-fi
-
-# Check if session is actually active
-AUTO_SESSION_ACTIVE=$(jq -r '.active // false' "$AUTO_SESSION_FILE" 2>/dev/null || echo "false")
-if [ "$AUTO_SESSION_ACTIVE" != "true" ]; then
-    silent_approve "Auto mode session not active" "session_inactive" '{"sessionActive":false}'
-fi
-
-# Update session file mtime to keep it fresh (touch it)
-touch "$AUTO_SESSION_FILE" 2>/dev/null
-
-# ============================================================================
-# CHECK FOR CUSTOM SUCCESS CRITERIA (auto-enables LLM eval)
-# ============================================================================
-
-HAS_CUSTOM_CRITERIA="false"
-LLM_EVAL_MODEL="opus"  # Default model for LLM evaluation (ultrathink)
-USER_GOAL=""
-SUCCESS_SUMMARY=""
-
-# Read success criteria from auto-mode.json
-CRITERIA_COUNT=$(jq -r '.successCriteria | length // 0' "$AUTO_SESSION_FILE" 2>/dev/null || echo "0")
-if [ "$CRITERIA_COUNT" -gt 2 ]; then
-    # More than default criteria (tasks_complete + acs_satisfied) = custom criteria
-    HAS_CUSTOM_CRITERIA="true"
-    REQUIRE_LLM_EVAL="true"  # Auto-enable LLM eval when custom criteria exist
-    log "Custom success criteria detected ($CRITERIA_COUNT), enabling LLM evaluation"
-fi
-
-# Check if any criterion has type=llm_evaluate
-HAS_LLM_CRITERION=$(jq -r '.successCriteria[]? | select(.type == "llm_evaluate") | .type' "$AUTO_SESSION_FILE" 2>/dev/null | head -1)
-if [ -n "$HAS_LLM_CRITERION" ]; then
-    REQUIRE_LLM_EVAL="true"
-    log "LLM evaluation criterion found, enabling LLM evaluation"
-fi
-
-# Read user goal and success summary for logging
-USER_GOAL=$(jq -r '.userGoal // ""' "$AUTO_SESSION_FILE" 2>/dev/null || echo "")
-SUCCESS_SUMMARY=$(jq -r '.successSummary // "All tasks and acceptance criteria complete"' "$AUTO_SESSION_FILE" 2>/dev/null)
-
-log "Config: TDD=$TDD_MODE, RequireTests=$REQUIRE_TESTS, RequireValidation=$REQUIRE_VALIDATION, RequireJudgeLLM=$REQUIRE_JUDGE_LLM, RequireLLMEval=$REQUIRE_LLM_EVAL, MaxTurns=$MAX_TURNS, MaxRetries=$MAX_RETRIES"
-log "Auto mode session active: $AUTO_SESSION_ACTIVE"
-log "User goal: $USER_GOAL"
-log "Success summary: $SUCCESS_SUMMARY"
-
-# ============================================================================
-# TURN COUNTER - HARD STOP after maxTurns (NEVER resets during session)
-# This is the PRIMARY limit on session length. Unlike retry counter which
-# resets when increments change, this counter NEVER resets during a session.
-# ============================================================================
-
-TURN_FILE="$STATE_DIR/.stop-auto-turns"
-CURRENT_TURN=1
-
-# Read and increment turn counter (atomic operation)
-if [ -f "$TURN_FILE" ]; then
-    STORED_TURN=$(cat "$TURN_FILE" 2>/dev/null || echo "0")
-    # Validate it's a number
-    if [[ "$STORED_TURN" =~ ^[0-9]+$ ]]; then
-        CURRENT_TURN=$((STORED_TURN + 1))
-    fi
-fi
-
-# Write new turn count (atomic via temp file)
-echo "$CURRENT_TURN" > "$TURN_FILE.tmp" 2>/dev/null && mv "$TURN_FILE.tmp" "$TURN_FILE" 2>/dev/null
-
-log "Turn counter: $CURRENT_TURN / $MAX_TURNS"
-
-# HARD STOP: If turn count reaches limit, end session immediately
-if [ "$CURRENT_TURN" -ge "$MAX_TURNS" ]; then
-    log "HARD STOP: Turn limit reached ($CURRENT_TURN >= $MAX_TURNS)"
-
-    # Clean up all state files
-    rm -f "$TURN_FILE" 2>/dev/null
-    rm -f "$STATE_DIR/.stop-auto-dedup" 2>/dev/null
-    rm -f "$STATE_DIR/.stop-auto-retry" 2>/dev/null
-    rm -f "$AUTO_SESSION_FILE" 2>/dev/null
-
-    # Get active increments for the message
-    ACTIVE_INCS=$(find "$INCREMENTS_DIR" -maxdepth 2 -name "metadata.json" \
-        -exec grep -l '"status"[[:space:]]*:[[:space:]]*"active\|"status"[[:space:]]*:[[:space:]]*"in-progress' {} \; 2>/dev/null \
-        | sed 's|.*/\([^/]*\)/metadata.json|\1|' | sort | tr '\n' ', ' | sed 's/,$//')
-
-    TURN_LIMIT_MSG="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-🛑 SESSION TURN LIMIT REACHED ($CURRENT_TURN/$MAX_TURNS)
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-The auto mode session has reached its maximum turn limit.
-This is a safety mechanism to prevent runaway sessions.
-
-📋 REMAINING WORK:
-   • Active increment(s): ${ACTIVE_INCS:-none}
-
-🔧 OPTIONS:
-   1. Start a new auto session: /sw:auto
-   2. Continue manually: /sw:do
-   3. Check status: /sw:progress
-
-💡 To increase turn limit, set in .specweave/config.json:
-   { \"auto\": { \"maxTurns\": 100 } }
-
-   Current limit: $MAX_TURNS turns
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-    # Log structured decision for turn limit
-    if type log_decision &>/dev/null; then
-        _turn_context=$(jq -n \
-            --argjson turnCurrent "$CURRENT_TURN" \
-            --argjson turnMax "$MAX_TURNS" \
-            --arg activeIncs "${ACTIVE_INCS:-none}" \
-            '{
-                sessionActive: true,
-                turn: {current: $turnCurrent, max: $turnMax},
-                increments: {active: ($activeIncs | split(", "))}
-            }')
-        log_decision "stop-auto" "approve" "turn_limit" "Turn limit reached: $CURRENT_TURN/$MAX_TURNS turns" "$_turn_context" "$(_get_duration_ms)"
-    fi
-
-    jq -n \
-        --arg decision "approve" \
-        --arg reason "Turn limit reached: $CURRENT_TURN/$MAX_TURNS turns" \
-        --arg msg "$TURN_LIMIT_MSG" \
-        '{decision: $decision, reason: $reason, systemMessage: $msg}'
-    exit 0
-fi
-
-# ============================================================================
-# DEDUPLICATION - Prevent feedback loops (Claude Code UI bug workaround)
-# ============================================================================
-
-DEDUP_FILE="$STATE_DIR/.stop-auto-dedup"
-DEDUP_WINDOW="${SPECWEAVE_STOP_HOOK_DEDUP:-5}"
-NOW=$(date +%s)
-
-if [ -f "$DEDUP_FILE" ]; then
-    LAST_FIRE=$(cat "$DEDUP_FILE" 2>/dev/null || echo "0")
-    ELAPSED=$((NOW - LAST_FIRE))
-    [ "$ELAPSED" -gt 3600 ] && rm -f "$DEDUP_FILE" 2>/dev/null
-    [ "$ELAPSED" -lt "$DEDUP_WINDOW" ] && silent_approve "Deduplicated (${ELAPSED}s < ${DEDUP_WINDOW}s)" "deduplicated" "$(jq -n --argjson elapsed "$ELAPSED" --argjson window "$DEDUP_WINDOW" '{elapsed:$elapsed,dedupWindow:$window}')"
-fi
-
-mkdir -p "$STATE_DIR" 2>/dev/null
-echo "$NOW" > "$DEDUP_FILE" 2>/dev/null
-
-# ============================================================================
-# BLOCK RETRY COUNTER - Track how many times we block on same increments
-# ============================================================================
-
-RETRY_FILE="$STATE_DIR/.stop-auto-retry"
-RETRY_COUNT=0
-MAX_RETRIES_BEFORE_ESCALATE="$MAX_RETRIES"  # From config (default: 20)
-
-# Read current retry state
-if [ -f "$RETRY_FILE" ]; then
-    RETRY_COUNT=$(jq -r '.count // 0' "$RETRY_FILE" 2>/dev/null || echo "0")
-    RETRY_INCS=$(jq -r '.increments // ""' "$RETRY_FILE" 2>/dev/null || echo "")
-fi
-
-# Function to update retry count (called when blocking)
-# Also tracks failure reasons for reflection
-update_retry_counter() {
-    local incs="$1"
-    local reason="$2"
-    local new_count=$((RETRY_COUNT + 1))
-
-    # Reset if different increments
-    if [ "$RETRY_INCS" != "$incs" ]; then
-        new_count=1
-    fi
-
-    # Get previous reasons for reflection
-    local prev_reasons=""
-    if [ -f "$RETRY_FILE" ]; then
-        prev_reasons=$(jq -r '.reasons // []' "$RETRY_FILE" 2>/dev/null || echo "[]")
-    fi
-
-    # Build updated retry state with history
-    jq -n \
-        --argjson count "$new_count" \
-        --arg increments "$incs" \
-        --arg timestamp "$(date -u +%Y-%m-%dT%H:%M:%SZ)" \
-        --arg currentReason "$reason" \
-        --argjson prevReasons "$prev_reasons" \
-        '{
-          count: $count,
-          increments: $increments,
-          lastUpdate: $timestamp,
-          currentReason: $currentReason,
-          reasons: ([$currentReason] + ($prevReasons | .[0:4]))
-        }' \
-        > "$RETRY_FILE" 2>/dev/null
-
-    echo "$new_count"
-}
-
-# Function to get previous failure reasons for reflection
-get_failure_history() {
-    if [ -f "$RETRY_FILE" ]; then
-        jq -r '.reasons // [] | .[] | "  - " + .' "$RETRY_FILE" 2>/dev/null || echo ""
-    fi
-}
-
-# Function to clear retry counter (called when work completes)
-clear_retry_counter() {
-    rm -f "$RETRY_FILE" 2>/dev/null
-}
-
-# Function to clear auto-mode session (called when all work completes)
-clear_auto_session() {
-    rm -f "$AUTO_SESSION_FILE" 2>/dev/null
-    rm -f "$TURN_FILE" 2>/dev/null  # Also clear turn counter on session end
-    log "Auto-mode session cleared (including turn counter)"
-}
-
-# ============================================================================
-# CHECK FOR PARALLEL SESSION
-# ============================================================================
-
-PARALLEL_SESSION="$STATE_DIR/parallel/session.json"
-
-if [ -f "$PARALLEL_SESSION" ]; then
-    SESSION_STATUS=$(jq -r '.status // "unknown"' "$PARALLEL_SESSION" 2>/dev/null || echo "unknown")
-
-    if [ "$SESSION_STATUS" = "active" ]; then
-        PENDING_AGENTS=$(jq '[.agents[] | select(.status != "completed" and .status != "failed" and .status != "cancelled")] | length' "$PARALLEL_SESSION" 2>/dev/null || echo "0")
-
-        if [ "$PENDING_AGENTS" -gt 0 ]; then
-            AGENT_LIST=$(jq -r '[.agents[] | select(.status != "completed" and .status != "failed" and .status != "cancelled") | "\(.domain):\(.status)"] | join(", ")' "$PARALLEL_SESSION" 2>/dev/null || echo "unknown")
-
-            MSG="🔄 $PENDING_AGENTS parallel agent(s) running: $AGENT_LIST → wait for completion"
-            log "BLOCK: Parallel agents running: $AGENT_LIST"
-
-            jq -n \
-                --arg decision "block" \
-                --arg reason "Parallel agents still running" \
-                --arg msg "$MSG" \
-                '{decision: $decision, reason: $reason, systemMessage: $msg}'
-            exit 0
-        fi
-    fi
-fi
-
-# ============================================================================
-# HELPER: Count pending tasks in tasks.md
-# Returns: number of tasks with "**Status**: [ ] pending"
-# ============================================================================
-count_pending_tasks() {
-    local inc_id="$1"
-    local tasks_file="$INCREMENTS_DIR/$inc_id/tasks.md"
-
-    if [ ! -f "$tasks_file" ]; then
-        echo "0"
-        return
-    fi
-
-    # Count tasks with **Status**: [ ] pending (case insensitive)
-    grep -ciE '\*\*Status\*\*:\s*\[\s*\]\s*pending' "$tasks_file" 2>/dev/null || echo "0"
-}
-
-# ============================================================================
-# HELPER: Check if increment has all ACs completed
-# Returns: 0 if all complete, count of open ACs otherwise
-# ============================================================================
-count_open_acs() {
-    local inc_id="$1"
-    local spec_file="$INCREMENTS_DIR/$inc_id/spec.md"
-
-    if [ ! -f "$spec_file" ]; then
-        echo "0"
-        return
-    fi
-
-    # Count unchecked ACs: - [ ] **AC-
-    # Note: grep -c returns exit code 1 when count is 0, so use a temp var
-    local count
-    count=$(grep -c '^- \[ \] \*\*AC-' "$spec_file" 2>/dev/null) || count=0
-    echo "$count"
-}
-
-# ============================================================================
-# HELPER: Run tests and return result
-# Returns: "pass" or "fail:reason"
-# ============================================================================
-run_tests() {
-    local inc_id="$1"
-
-    # Try to find and run tests
-    # Priority: npm test, yarn test, bun test, pytest, go test
-
-    if [ -f "$PROJECT_ROOT/package.json" ]; then
-        # Check if test script exists
-        local has_test=$(jq -r '.scripts.test // ""' "$PROJECT_ROOT/package.json" 2>/dev/null)
-        if [ -n "$has_test" ] && [ "$has_test" != "null" ]; then
-            log "Running: npm test"
-            # Run tests with timeout, capture exit code
-            cd "$PROJECT_ROOT" && timeout 300 npm test >/dev/null 2>&1
-            local exit_code=$?
-            if [ $exit_code -eq 0 ]; then
-                echo "pass"
-            else
-                echo "fail:npm test failed (exit $exit_code)"
-            fi
-            return
-        fi
-    fi
-
-    # Check for Python tests
-    if [ -f "$PROJECT_ROOT/pytest.ini" ] || [ -d "$PROJECT_ROOT/tests" ]; then
-        if command -v pytest >/dev/null 2>&1; then
-            log "Running: pytest"
-            cd "$PROJECT_ROOT" && timeout 300 pytest -q >/dev/null 2>&1
-            local exit_code=$?
-            if [ $exit_code -eq 0 ]; then
-                echo "pass"
-            else
-                echo "fail:pytest failed (exit $exit_code)"
-            fi
-            return
-        fi
-    fi
-
-    # Check for Go tests
-    if [ -f "$PROJECT_ROOT/go.mod" ]; then
-        if command -v go >/dev/null 2>&1; then
-            log "Running: go test"
-            cd "$PROJECT_ROOT" && timeout 300 go test ./... >/dev/null 2>&1
-            local exit_code=$?
-            if [ $exit_code -eq 0 ]; then
-                echo "pass"
-            else
-                echo "fail:go test failed (exit $exit_code)"
-            fi
-            return
-        fi
-    fi
-
-    # No tests found - pass by default (user should configure tests)
-    echo "pass:no-tests-configured"
-}
-
-# ============================================================================
-# HELPER: Validate increment completion
-# Returns: "valid" or "invalid:reason"
-# ============================================================================
-validate_increment() {
-    local inc_id="$1"
-
-    # Check pending tasks
-    local pending=$(count_pending_tasks "$inc_id")
-    if [ "$pending" -gt 0 ]; then
-        echo "invalid:$pending tasks still pending"
-        return
-    fi
-
-    # Check open ACs
-    local open_acs=$(count_open_acs "$inc_id")
-    if [ "$open_acs" -gt 0 ]; then
-        echo "invalid:$open_acs acceptance criteria still open"
-        return
-    fi
-
-    # If TDD mode or requireTests, run tests
-    if [ "$TDD_MODE" = "true" ] || [ "$REQUIRE_TESTS" = "true" ]; then
-        local test_result=$(run_tests "$inc_id")
-        if [[ "$test_result" == fail:* ]]; then
-            local reason="${test_result#fail:}"
-            echo "invalid:Tests failed - $reason"
-            return
-        fi
-        log "Tests passed for $inc_id"
-    fi
-
-    # If requireLLMEval, run LLM-based completion evaluation
-    if [ "$REQUIRE_LLM_EVAL" = "true" ]; then
-        log "Running LLM completion evaluation for $inc_id..."
-
-        # Check if specweave CLI is available
-        if command -v specweave >/dev/null 2>&1; then
-            local eval_result
-            eval_result=$(cd "$PROJECT_ROOT" && timeout 60 specweave evaluate-completion "$inc_id" --model opus --silent 2>/dev/null)
-            local eval_exit=$?
-
-            if [ "$eval_exit" -eq 0 ]; then
-                log "LLM evaluation: COMPLETE for $inc_id"
-            else
-                # Parse the reason from JSON if possible
-                local eval_reason=""
-                if [ -n "$eval_result" ]; then
-                    eval_reason=$(echo "$eval_result" | jq -r '.overallReason // "Unknown reason"' 2>/dev/null || echo "Unknown reason")
-                else
-                    eval_reason="LLM evaluation returned incomplete"
-                fi
-                log "LLM evaluation: NOT COMPLETE - $eval_reason"
-                echo "invalid:LLM evaluation incomplete - $eval_reason"
-                return
-            fi
-        else
-            log "WARN: specweave CLI not available, skipping LLM evaluation"
-        fi
-    fi
-
-    echo "valid"
-}
-
-# ============================================================================
-# HELPER: Auto-close increment (update status to completed)
-# This triggers StatusChangeSyncTrigger → external sync!
-# ============================================================================
-auto_close_increment() {
-    local inc_id="$1"
-    local metadata_file="$INCREMENTS_DIR/$inc_id/metadata.json"
-
-    if [ ! -f "$metadata_file" ]; then
-        log "ERROR: metadata.json not found for $inc_id"
-        return 1
-    fi
-
-    log "AUTO-CLOSE: Updating $inc_id to completed"
-
-    # Use specweave CLI if available (triggers proper hooks and sync)
-    if command -v specweave >/dev/null 2>&1; then
-        cd "$PROJECT_ROOT" && specweave complete "$inc_id" --yes --silent 2>/dev/null
-        if [ $? -eq 0 ]; then
-            log "AUTO-CLOSE: specweave complete succeeded for $inc_id"
-            return 0
-        fi
-        log "WARN: specweave complete failed, falling back to direct update"
-    fi
-
-    # Fallback: Direct metadata update (triggers StatusChangeSyncTrigger via node)
-    # This is less ideal but ensures we don't get stuck
-    local now=$(date -u +%Y-%m-%dT%H:%M:%SZ)
-    local temp_file=$(mktemp)
-
-    jq --arg now "$now" '.status = "completed" | .completedAt = $now | .updated = $now' \
-        "$metadata_file" > "$temp_file" && mv "$temp_file" "$metadata_file"
-
-    if [ $? -eq 0 ]; then
-        log "AUTO-CLOSE: Direct metadata update succeeded for $inc_id"
-
-        # Trigger sync via node helper if available
-        local sync_helper="$PROJECT_ROOT/node_modules/specweave/dist/hooks/trigger-sync.js"
-        if [ -f "$sync_helper" ]; then
-            node "$sync_helper" "$inc_id" 2>/dev/null &
-        fi
-        return 0
-    else
-        log "ERROR: Failed to update metadata for $inc_id"
-        return 1
-    fi
-}
-
-# ============================================================================
-# MAIN: Find active increments and validate each
-# ============================================================================
-
-# Find all active increments
-ACTIVE_METADATA_FILES=$(find "$INCREMENTS_DIR" -maxdepth 2 -name "metadata.json" \
-    -exec grep -l '"status"[[:space:]]*:[[:space:]]*"active\|"status"[[:space:]]*:[[:space:]]*"in-progress' {} \; 2>/dev/null)
-
-# Arrays to track status
-INCOMPLETE_INCS=""
-READY_TO_CLOSE=""
-VALIDATION_ERRORS=""
-
-# Cache for first incomplete increment (avoids redundant calculation later)
-FIRST_INC_PENDING_CACHED=""
-FIRST_INC_ACS_CACHED=""
-FIRST_INC_CACHED=""
-
-for meta_file in $ACTIVE_METADATA_FILES; do
-    # Extract increment ID from path
-    inc_id=$(echo "$meta_file" | sed 's|.*/\([^/]*\)/metadata.json|\1|')
-
-    log "Checking increment: $inc_id"
-
-    if [ "$SKIP_QUALITY_GATES" = "true" ]; then
-        # Skip validation, just check task status
-        pending=$(count_pending_tasks "$inc_id")
-        open_acs=$(count_open_acs "$inc_id")
-        if [ "$pending" -eq 0 ]; then
-            log "SKIP_QUALITY_GATES: $inc_id has no pending tasks, marking ready"
-            READY_TO_CLOSE="$READY_TO_CLOSE $inc_id"
-        else
-            INCOMPLETE_INCS="$INCOMPLETE_INCS $inc_id"
-            VALIDATION_ERRORS="$VALIDATION_ERRORS|$inc_id:$pending tasks pending"
-            # Cache first incomplete increment's stats
-            if [ -z "$FIRST_INC_CACHED" ]; then
-                FIRST_INC_CACHED="$inc_id"
-                FIRST_INC_PENDING_CACHED="$pending"
-                FIRST_INC_ACS_CACHED="$open_acs"
-            fi
-        fi
-    else
-        # Full validation - get counts first for caching
-        pending=$(count_pending_tasks "$inc_id")
-        open_acs=$(count_open_acs "$inc_id")
-        result=$(validate_increment "$inc_id")
-
-        if [ "$result" = "valid" ]; then
-            log "VALID: $inc_id ready to close"
-            READY_TO_CLOSE="$READY_TO_CLOSE $inc_id"
-        else
-            reason="${result#invalid:}"
-            log "INVALID: $inc_id - $reason"
-            INCOMPLETE_INCS="$INCOMPLETE_INCS $inc_id"
-            VALIDATION_ERRORS="$VALIDATION_ERRORS|$inc_id:$reason"
-            # Cache first incomplete increment's stats
-            if [ -z "$FIRST_INC_CACHED" ]; then
-                FIRST_INC_CACHED="$inc_id"
-                FIRST_INC_PENDING_CACHED="$pending"
-                FIRST_INC_ACS_CACHED="$open_acs"
-            fi
-        fi
-    fi
-done
-
-# ============================================================================
-# AUTO-CLOSE: Close increments that passed validation
-# This is the KEY fix - actually transition status to trigger sync!
-# ============================================================================
-
-CLOSED_COUNT=0
-for inc_id in $READY_TO_CLOSE; do
-    [ -z "$inc_id" ] && continue
-
-    auto_close_increment "$inc_id"
-    if [ $? -eq 0 ]; then
-        CLOSED_COUNT=$((CLOSED_COUNT + 1))
-        log "CLOSED: $inc_id"
-    fi
-done
-
-# ============================================================================
-# SKILL VALIDATION: Run domain-specific validation if skills were activated
-# Only runs in auto mode when skills/agents were used during session
-# ============================================================================
-
-SKILL_ACTIVATIONS_FILE="$STATE_DIR/skill-activations.json"
-SKILL_VALIDATION_FAILED="false"
-VALIDATION_SUMMARY=""  # Track what validations were run
-
-if [ -f "$SKILL_ACTIVATIONS_FILE" ]; then
-    # Skills were activated during this session
-    ACTIVATED_DOMAINS=$(jq -r '.domains | join(", ")' "$SKILL_ACTIVATIONS_FILE" 2>/dev/null || echo "")
-    ACTIVATION_COUNT=$(jq -r '.activations | length' "$SKILL_ACTIVATIONS_FILE" 2>/dev/null || echo "0")
-
-    if [ -n "$ACTIVATED_DOMAINS" ] && [ "$ACTIVATED_DOMAINS" != "null" ]; then
-        log "SKILL VALIDATION: Activated domains: $ACTIVATED_DOMAINS"
-
-        # Run domain-specific validation
-        VALIDATION_SCRIPT="$PROJECT_ROOT/plugins/specweave/hooks/validate-skill-activations.sh"
-
-        # Try installed location first, then package location
-        if [ ! -f "$VALIDATION_SCRIPT" ]; then
-            VALIDATION_SCRIPT="$(dirname "$0")/validate-skill-activations.sh"
-        fi
-
-        if [ -f "$VALIDATION_SCRIPT" ] && [ -x "$VALIDATION_SCRIPT" ]; then
-            log "Running skill validation: $VALIDATION_SCRIPT"
-
-            SKILL_VALIDATION_START=$(date +%s)
-            if ! PROJECT_ROOT="$PROJECT_ROOT" "$VALIDATION_SCRIPT" 2>&1 | tee -a "$LOG_FILE"; then
-                SKILL_VALIDATION_FAILED="true"
-                SKILL_VALIDATION_END=$(date +%s)
-                SKILL_VALIDATION_DURATION=$((SKILL_VALIDATION_END - SKILL_VALIDATION_START))
-                log "SKILL VALIDATION: Failed - blocking completion"
-                VALIDATION_ERRORS="$VALIDATION_ERRORS|skill_validation:Domain validation failed for: $ACTIVATED_DOMAINS"
-                VALIDATION_SUMMARY="${VALIDATION_SUMMARY}
-  ❌ Skill Validation ($ACTIVATED_DOMAINS): FAILED in ${SKILL_VALIDATION_DURATION}s"
-            else
-                SKILL_VALIDATION_END=$(date +%s)
-                SKILL_VALIDATION_DURATION=$((SKILL_VALIDATION_END - SKILL_VALIDATION_START))
-                log "SKILL VALIDATION: Passed"
-                VALIDATION_SUMMARY="${VALIDATION_SUMMARY}
-  ✅ Skill Validation ($ACTIVATED_DOMAINS): PASSED in ${SKILL_VALIDATION_DURATION}s"
-            fi
-        else
-            log "SKILL VALIDATION: Script not found, skipping"
-            VALIDATION_SUMMARY="${VALIDATION_SUMMARY}
-  ⏭️  Skill Validation: Skipped (no validator)"
-        fi
-    fi
-else
-    # No skills activated - note this in log
-    log "SKILL VALIDATION: No skills activated this session"
-fi
-
-# Add increment validation summary
-if [ -n "$READY_TO_CLOSE" ]; then
-    for inc in $READY_TO_CLOSE; do
-        [ -z "$inc" ] && continue
-        VALIDATION_SUMMARY="${VALIDATION_SUMMARY}
-  ✅ Increment $inc: All tasks complete, ACs satisfied"
-    done
-fi
-
-if [ -n "$INCOMPLETE_INCS" ]; then
-    for inc in $INCOMPLETE_INCS; do
-        [ -z "$inc" ] && continue
-        VALIDATION_SUMMARY="${VALIDATION_SUMMARY}
-  ⏳ Increment $inc: Work remaining"
-    done
-fi
-
-# Log validation summary
-if [ -n "$VALIDATION_SUMMARY" ]; then
-    log "VALIDATION SUMMARY:$VALIDATION_SUMMARY"
-fi
-
-# ============================================================================
-# DECISION: Continue or Complete
-# ============================================================================
-
-# Recount after auto-close
-REMAINING_COUNT=$(find "$INCREMENTS_DIR" -maxdepth 2 -name "metadata.json" \
-    -exec grep -l '"status"[[:space:]]*:[[:space:]]*"active\|"status"[[:space:]]*:[[:space:]]*"in-progress' {} \; 2>/dev/null | wc -l | tr -d ' ')
-
-if [ "$REMAINING_COUNT" -eq 0 ] && [ "$SKILL_VALIDATION_FAILED" != "true" ]; then
-    # All increments closed AND skill validation passed - approve exit
-    rm -f "$DEDUP_FILE" 2>/dev/null
-    clear_retry_counter
-    clear_auto_session  # Clear session marker - work is done!
-
-    # Build context for logging
-    _complete_context=$(jq -n \
-        --argjson turnCurrent "$CURRENT_TURN" \
-        --argjson turnMax "$MAX_TURNS" \
-        --argjson closedCount "$CLOSED_COUNT" \
-        '{
-            sessionActive: false,
-            turn: {current: $turnCurrent, max: $turnMax},
-            closedIncrements: $closedCount
-        }')
-
-    if [ "$CLOSED_COUNT" -gt 0 ]; then
-        log "APPROVE: Auto-closed $CLOSED_COUNT increment(s), all work complete"
-        # Show completion message to user
-        COMPLETE_MSG="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-✅ AUTO MODE COMPLETE (Turn $CURRENT_TURN/$MAX_TURNS)
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-🎉 Auto-closed $CLOSED_COUNT increment(s) successfully!
-
-📊 Session Summary:
-   • Turns used: $CURRENT_TURN
-   • Increments completed: $CLOSED_COUNT
-
-💡 Next steps:
-   • Start new work: /sw:increment \"feature name\"
-   • View completed: /sw:status
-   • Review logs: specweave decision-log --since 1h
-
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-        loud_approve "Auto-closed $CLOSED_COUNT increment(s)" "all_complete" "$_complete_context" "$COMPLETE_MSG"
-    else
-        # No active increments - show brief message
-        NO_WORK_MSG="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-✅ NO ACTIVE INCREMENTS
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-No active increments to work on. Session ended.
-
-💡 Start new work: /sw:increment \"feature name\"
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-        loud_approve "No active increments" "all_complete" "$_complete_context" "$NO_WORK_MSG"
-    fi
-fi
-
-# ============================================================================
-# KEY FIX: APPROVE if all tasks complete even if increments still "active"
-# This handles: auto-close failed, user prefers manual close, etc.
-# ============================================================================
-
-# Count how many active increments have incomplete work
-INCOMPLETE_COUNT=$(echo "$INCOMPLETE_INCS" | tr -s ' ' | sed 's/^ //' | grep -v '^$' | wc -w | tr -d ' ')
-
-if [ "$INCOMPLETE_COUNT" -eq 0 ] && [ "$SKILL_VALIDATION_FAILED" != "true" ]; then
-    # All active increments have complete tasks/ACs - approve even if still "active" status
-    rm -f "$DEDUP_FILE" 2>/dev/null
-    clear_retry_counter
-    clear_auto_session  # Clear session marker - work is done!
-
-    log "APPROVE: All tasks complete in active increments (manual close pending)"
-
-    # Log structured decision
-    if type log_decision &>/dev/null; then
-        _work_complete_context=$(jq -n \
-            --argjson turnCurrent "$CURRENT_TURN" \
-            --argjson turnMax "$MAX_TURNS" \
-            --arg readyToClose "$READY_TO_CLOSE" \
-            '{sessionActive: false, turn: {current: $turnCurrent, max: $turnMax}, increments: {active: [], readyToClose: ($readyToClose | split(" ") | map(select(length > 0)))}}')
-        log_decision "stop-auto" "approve" "work_complete" "All tasks complete - increments ready for manual close" "$_work_complete_context" "$(_get_duration_ms)"
-    fi
-
-    # Show message about manual close needed
-    CLOSE_MSG="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-✅ ALL TASKS COMPLETE - Session can end
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-All work is done! Increments need manual closing:
-
-$READY_TO_CLOSE
-
-To close: /sw:done <increment-id>
-
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-    jq -n \
-        --arg decision "approve" \
-        --arg reason "All tasks complete - increments ready for manual close" \
-        --arg msg "$CLOSE_MSG" \
-        '{decision: $decision, reason: $reason, systemMessage: $msg}'
-    exit 0
-fi
-
-# Handle skill validation failure (block even if tasks are done)
-if [ "$SKILL_VALIDATION_FAILED" = "true" ]; then
-    log "BLOCK: Skill validation failed - fix issues before completion"
-fi
-
-# Work remains - show what's blocking
-# IMPORTANT: Sort to ensure deterministic order (prevents retry counter resets)
-REMAINING_INCS=$(find "$INCREMENTS_DIR" -maxdepth 2 -name "metadata.json" \
-    -exec grep -l '"status"[[:space:]]*:[[:space:]]*"active\|"status"[[:space:]]*:[[:space:]]*"in-progress' {} \; 2>/dev/null \
-    | sed 's|.*/\([^/]*\)/metadata.json|\1|' | sort | tr '\n' ', ' | sed 's/,$//')
-
-# Build reason string for retry tracking
-BLOCK_REASON="${VALIDATION_ERRORS:-tasks_or_acs_pending}"
-
-# Update retry counter with reason
-CURRENT_RETRY=$(update_retry_counter "$REMAINING_INCS" "$BLOCK_REASON")
-log "Retry count: $CURRENT_RETRY for increments: $REMAINING_INCS"
-
-# Get first increment ID for commands
-FIRST_INC=$(echo "$REMAINING_INCS" | cut -d',' -f1 | tr -d ' ')
-
-# ============================================================================
-# DETECT PROJECT CAPABILITIES (conditional steps)
-# ============================================================================
-
-HAS_TESTS="false"
-TEST_CMD=""
-
-# Check for test capability
-if [ -f "$PROJECT_ROOT/package.json" ]; then
-    TEST_SCRIPT=$(jq -r '.scripts.test // ""' "$PROJECT_ROOT/package.json" 2>/dev/null)
-    if [ -n "$TEST_SCRIPT" ] && [ "$TEST_SCRIPT" != "null" ]; then
-        HAS_TESTS="true"
-        TEST_CMD="npm test"
-    fi
-elif [ -f "$PROJECT_ROOT/pytest.ini" ] || [ -d "$PROJECT_ROOT/tests" ]; then
-    HAS_TESTS="true"
-    TEST_CMD="pytest"
-elif [ -f "$PROJECT_ROOT/go.mod" ]; then
-    HAS_TESTS="true"
-    TEST_CMD="go test ./..."
-fi
-
-# Use configured test command if provided
-[ -n "$TEST_COMMAND" ] && TEST_CMD="$TEST_COMMAND" && HAS_TESTS="true"
-
-# ============================================================================
-# BUILD MESSAGE WITH REFLECTION ON PREVIOUS FAILURES
-# ============================================================================
-
-MSG=""
-
-# Add validation summary at the top if any validations were run
-if [ -n "$VALIDATION_SUMMARY" ]; then
-    MSG="━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-📊 VALIDATION SUMMARY
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-$VALIDATION_SUMMARY
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-"
-fi
-
-# Add closed count if any
-if [ "$CLOSED_COUNT" -gt 0 ]; then
-    MSG="${MSG}✅ Auto-closed $CLOSED_COUNT increment(s)
-
-"
-fi
-
-# Build header based on retry count
-if [ "$CURRENT_RETRY" -ge "$MAX_RETRIES_BEFORE_ESCALATE" ]; then
-    MSG="${MSG}🚨 STUCK SESSION (attempt $CURRENT_RETRY/$MAX_RETRIES_BEFORE_ESCALATE)
-
-⚠️  This session has failed to complete $CURRENT_RETRY times.
-    Consider: pausing (/sw:pause $FIRST_INC) or abandoning (/sw:abandon $FIRST_INC)
-
-"
-fi
-
-MSG="${MSG}🔄 $REMAINING_COUNT increment(s) need work: $REMAINING_INCS
-📊 Session: turn $CURRENT_TURN/$MAX_TURNS | stuck retry $CURRENT_RETRY/$MAX_RETRIES_BEFORE_ESCALATE"
-
-# Add validation errors if any
-if [ -n "$VALIDATION_ERRORS" ]; then
-    error_details=$(echo "$VALIDATION_ERRORS" | tr '|' '\n' | grep -v '^$' | head -5 | while read line; do
-        echo "   • $line"
-    done)
-    MSG="$MSG
-
-📋 Current blocking issues:
-$error_details"
-fi
-
-# ============================================================================
-# REFLECTION: Why did previous attempts fail?
-# ============================================================================
-
-if [ "$CURRENT_RETRY" -gt 1 ]; then
-    FAILURE_HISTORY=$(get_failure_history)
-    if [ -n "$FAILURE_HISTORY" ]; then
-        MSG="$MSG
-
-🔍 REFLECT: Why previous attempts failed:
-$FAILURE_HISTORY
-
-💡 THINK: What can you do differently this time?
-   • Is there a different approach to fix the issue?
-   • Are you stuck in a loop doing the same thing?
-   • Should you ask the user for help?"
-    fi
-fi
-
-# Add mode indicators
-if [ "$TDD_MODE" = "true" ]; then
-    MSG="🔴 TDD MODE | $MSG"
-elif [ "$REQUIRE_TESTS" = "true" ]; then
-    MSG="🧪 TEST MODE | $MSG"
-fi
-
-# ============================================================================
-# BUILD CONDITIONAL COMPLETION STEPS
-# ============================================================================
-
-STEP_NUM=1
-STEPS=""
-
-# Step 1: Complete tasks (always required)
-STEPS="${STEPS}
-${STEP_NUM}️⃣  COMPLETE ALL TASKS
-    → Run: /sw:do
-    → Mark all tasks [x] completed in tasks.md
-    → Mark all ACs [x] completed in spec.md"
-STEP_NUM=$((STEP_NUM + 1))
-
-# Step 2: Run tests (if available and required)
-if [ "$HAS_TESTS" = "true" ] && { [ "$REQUIRE_TESTS" = "true" ] || [ "$TDD_MODE" = "true" ]; }; then
-    STEPS="${STEPS}
-
-${STEP_NUM}️⃣  RUN TESTS (REQUIRED by config)
-    → Run: $TEST_CMD
-    → ALL tests MUST pass before proceeding
-    → If tests fail, FIX them and re-run"
-    STEP_NUM=$((STEP_NUM + 1))
-elif [ "$HAS_TESTS" = "true" ]; then
-    STEPS="${STEPS}
-
-${STEP_NUM}️⃣  RUN TESTS (recommended)
-    → Run: $TEST_CMD
-    → Verify your changes work correctly"
-    STEP_NUM=$((STEP_NUM + 1))
-fi
-
-# Step 3: Validate (if required by config)
-if [ "$REQUIRE_VALIDATION" = "true" ]; then
-    STEPS="${STEPS}
-
-${STEP_NUM}️⃣  VALIDATE QUALITY GATES (REQUIRED)
-    → Run: /sw:validate $FIRST_INC
-    → Review and fix any blocking issues"
-    STEP_NUM=$((STEP_NUM + 1))
-fi
-
-# Step 4: Judge LLM (if required by config)
-if [ "$REQUIRE_JUDGE_LLM" = "true" ]; then
-    STEPS="${STEPS}
-
-${STEP_NUM}️⃣  AI QUALITY VERIFICATION (REQUIRED)
-    → Run: /sw:judge-llm $FIRST_INC
-    → Must pass before closing"
-    STEP_NUM=$((STEP_NUM + 1))
-fi
-
-# Final step: Close increment (always required)
-STEPS="${STEPS}
-
-${STEP_NUM}️⃣  CLOSE INCREMENT (FINAL STEP)
-    → Run: /sw:done $FIRST_INC
-    → This validates gates and closes properly"
-
-MSG="$MSG
-
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-📋 COMPLETION STEPS for $FIRST_INC:
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-$STEPS
-
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-🎯 Execute these steps IN ORDER, then session will complete automatically.
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-# ============================================================================
-# CIRCUIT BREAKER: Prevent infinite loops after MAX_RETRIES
-# NOTE: Uses -ge (>=) to trigger AT the limit, not after it
-# ============================================================================
-
-if [ "$CURRENT_RETRY" -ge "$MAX_RETRIES_BEFORE_ESCALATE" ]; then
-    log "CIRCUIT BREAKER: Reached $MAX_RETRIES_BEFORE_ESCALATE stuck retries (on same work), approving to break loop"
-
-    CIRCUIT_MSG="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-🛑 STUCK SESSION BREAKER (retry $CURRENT_RETRY/$MAX_RETRIES_BEFORE_ESCALATE on same work)
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-⚠️  Stop hook has fired $CURRENT_RETRY times on the SAME incomplete work.
-    This suggests the session is stuck and not making progress.
-
-📋 REMAINING WORK:
-   • $REMAINING_COUNT active increment(s): $REMAINING_INCS
-
-🔧 RECOMMENDED ACTIONS:
-   1. Run /sw:status to see what's incomplete
-   2. Run /sw:do to complete remaining tasks
-   3. Run /sw:done $FIRST_INC when ready to close
-
-💡 To increase stuck retry limit, set in .specweave/config.json:
-   { \"auto\": { \"maxRetries\": 50 } }
-
-   NOTE: maxRetries tracks stuck retries on SAME work.
-   For total session limit, use maxTurns instead.
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-    # Clear retry counter and auto session to prevent immediate re-triggering
-    clear_retry_counter
-    clear_auto_session  # End auto mode on circuit breaker
-
-    # Log structured decision
-    if type log_decision &>/dev/null; then
-        _circuit_context=$(jq -n \
-            --argjson turnCurrent "$CURRENT_TURN" \
-            --argjson turnMax "$MAX_TURNS" \
-            --argjson retryCurrent "$CURRENT_RETRY" \
-            --argjson retryMax "$MAX_RETRIES_BEFORE_ESCALATE" \
-            --arg activeIncs "$REMAINING_INCS" \
-            '{sessionActive: false, turn: {current: $turnCurrent, max: $turnMax}, retry: {current: $retryCurrent, max: $retryMax, stuck: true}, increments: {active: ($activeIncs | split(", ") | map(select(length > 0)))}}')
-        log_decision "stop-auto" "approve" "retry_limit" "Stuck session: $CURRENT_RETRY retries on same incomplete work" "$_circuit_context" "$(_get_duration_ms)"
-    fi
-
-    jq -n \
-        --arg decision "approve" \
-        --arg reason "Stuck session: $CURRENT_RETRY retries on same incomplete work" \
-        --arg msg "$CIRCUIT_MSG" \
-        '{decision: $decision, reason: $reason, systemMessage: $msg}'
-    exit 0
-fi
-
-# ============================================================================
-# ESCAPE GUIDANCE: Help user understand when to pivot vs continue
-# Adds guidance for new functionality/bug fixes that don't fit current increment
-# ============================================================================
-
-# Build escape options section for the message
-ESCAPE_OPTIONS="
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-🔄 WORKING ON SOMETHING DIFFERENT? (New functionality, bug fix, or urgent work)
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-If the current increment ($FIRST_INC) doesn't match what you need to work on:
-
-📋 OPTION 1: Create a NEW increment for new functionality
-   → Run: /sw:cancel-auto (stop auto mode first)
-   → Then: /sw:increment \"your new feature or bug fix description\"
-   → This will involve PM and Architect roles to properly plan
-
-📋 OPTION 2: Pause current work and start something new
-   → Run: /sw:pause $FIRST_INC
-   → Then: /sw:increment \"urgent bug fix\" --type=bug
-   → Resume later: /sw:resume $FIRST_INC
-
-📋 OPTION 3: Emergency quick fix (skip planning)
-   → Run: /sw:cancel-auto
-   → Work directly without increment tracking
-   → Better for tiny fixes that don't need tracking
-
-💡 TIP: If you're stuck in a loop, the RIGHT answer is often to:
-   1. Cancel auto mode: /sw:cancel-auto
-   2. Think about what you actually need
-   3. Start fresh with a proper increment
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-# Show escape options after 3+ retries or when turn count is high
-if [ "$CURRENT_RETRY" -ge 3 ] || [ "$CURRENT_TURN" -ge $((MAX_TURNS / 2)) ]; then
-    log "Adding escape guidance (retry=$CURRENT_RETRY, turn=$CURRENT_TURN)"
-    MSG="$MSG
-$ESCAPE_OPTIONS"
-fi
-
-# ============================================================================
-# BUILD CLEAR SUCCESS CRITERIA FOR THE REASON FIELD
-# This is what shows in Claude Code UI - must be ACTIONABLE
-# ============================================================================
-
-# Extract first blocking issue
-FIRST_BLOCKER=""
-if [ -n "$VALIDATION_ERRORS" ]; then
-    FIRST_BLOCKER=$(echo "$VALIDATION_ERRORS" | tr '|' '\n' | grep -v '^$' | head -1)
-fi
-
-# Build success criteria string with MANDATORY instructions
-SUCCESS_CRITERIA=""
-NEXT_COMMAND=""
-
-# Use cached values if available, otherwise calculate (fallback)
-if [ -n "$FIRST_INC_CACHED" ] && [ "$FIRST_INC_CACHED" = "$FIRST_INC" ]; then
-    pending_count="$FIRST_INC_PENDING_CACHED"
-    open_acs="$FIRST_INC_ACS_CACHED"
-else
-    pending_count=$(count_pending_tasks "$FIRST_INC")
-    open_acs=$(count_open_acs "$FIRST_INC")
-fi
-
-if [ -n "$FIRST_BLOCKER" ]; then
-    SUCCESS_CRITERIA="FIX: $FIRST_BLOCKER"
-    NEXT_COMMAND="/sw:do"
-elif [ "$pending_count" -gt 0 ]; then
-    SUCCESS_CRITERIA="MUST COMPLETE $pending_count pending task(s)"
-    NEXT_COMMAND="/sw:do"
-elif [ "$open_acs" -gt 0 ]; then
-    SUCCESS_CRITERIA="MUST SATISFY $open_acs open AC(s)"
-    NEXT_COMMAND="/sw:do"
-else
-    # All tasks and ACs done - MUST close now!
-    SUCCESS_CRITERIA="ALL DONE → MUST RUN /sw:done $FIRST_INC NOW"
-    NEXT_COMMAND="/sw:done $FIRST_INC"
-fi
-
-log "BLOCK: $REMAINING_COUNT increment(s) remaining - turn $CURRENT_TURN/$MAX_TURNS, retry $CURRENT_RETRY - $SUCCESS_CRITERIA"
-
-# Build actionable reason for Claude Code UI - this is the INSTRUCTION
-# Show turn/maxTurns (total session limit) in reason for visibility
-if [ "$pending_count" -eq 0 ] && [ "$open_acs" -eq 0 ]; then
-    # Everything complete - direct to close
-    ACTIONABLE_REASON="✅ [turn $CURRENT_TURN/$MAX_TURNS] ALL COMPLETE! RUN NOW: /sw:done $FIRST_INC"
-else
-    # Work remains - show what and then remind to close
-    ACTIONABLE_REASON="⏳ [turn $CURRENT_TURN/$MAX_TURNS] $SUCCESS_CRITERIA → RUN: $NEXT_COMMAND → THEN: /sw:done $FIRST_INC"
-fi
-
-# Append mandatory instruction to systemMessage
-MSG="$MSG
-
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-🚨 MANDATORY INSTRUCTION (YOU MUST FOLLOW THIS):
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-
-1. COMPLETE ALL PENDING TASKS: Run /sw:do and finish every [ ] task
-2. SATISFY ALL ACCEPTANCE CRITERIA: Mark all ACs as [x] in spec.md
-3. CLOSE THE INCREMENT: Run /sw:done $FIRST_INC
-
-⚠️  DO NOT STOP until you have run /sw:done $FIRST_INC successfully!
-⚠️  Session: turn $CURRENT_TURN of $MAX_TURNS (hard limit)
-
-Current status: $pending_count pending task(s), $open_acs open AC(s)
-Next command to run: $NEXT_COMMAND
-━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
-
-# Log structured decision before blocking
-if type log_decision &>/dev/null; then
-    # Get previous reasons from retry file
-    _prev_reasons_json="[]"
-    if [ -f "$RETRY_FILE" ] && [ -s "$RETRY_FILE" ]; then
-        _tmp_reasons=$(jq -c '.reasons // []' "$RETRY_FILE" 2>/dev/null)
-        [ -n "$_tmp_reasons" ] && _prev_reasons_json="$_tmp_reasons"
-    fi
-
-    # Build blocked increments array
-    _blocked_json="[]"
-    if [ -n "$FIRST_INC" ]; then
-        _blocked_json=$(jq -n \
-            --arg id "$FIRST_INC" \
-            --argjson tasksPending "$pending_count" \
-            --argjson acsOpen "$open_acs" \
-            --arg reason "$SUCCESS_CRITERIA" \
-            '[{id: $id, tasksPending: $tasksPending, acsOpen: $acsOpen, reason: $reason}]')
-    fi
-
-    # Pre-compute stuck value (avoid subshell in jq args)
-    _stuck_val="false"
-    [ "$CURRENT_RETRY" -ge "$MAX_RETRIES_BEFORE_ESCALATE" ] && _stuck_val="true"
-
-    # Build full context JSON
-    _block_context=$(jq -n \
-        --argjson turnCurrent "$CURRENT_TURN" \
-        --argjson turnMax "$MAX_TURNS" \
-        --argjson retryCurrent "$CURRENT_RETRY" \
-        --argjson retryMax "$MAX_RETRIES_BEFORE_ESCALATE" \
-        --argjson stuck "$_stuck_val" \
-        --arg activeIncs "$REMAINING_INCS" \
-        --argjson blocked "$_blocked_json" \
-        --argjson previousReasons "$_prev_reasons_json" \
-        '{sessionActive: true, turn: {current: $turnCurrent, max: $turnMax}, retry: {current: $retryCurrent, max: $retryMax, stuck: $stuck}, increments: {active: ($activeIncs | split(", ") | map(select(length > 0))), blocked: $blocked}, previousReasons: $previousReasons}')
-
-    log_decision "stop-auto" "block" "work_remaining" "$SUCCESS_CRITERIA" "$_block_context" "$(_get_duration_ms)"
-fi
-
-jq -n \
-    --arg decision "block" \
-    --arg reason "$ACTIONABLE_REASON" \
-    --arg msg "$MSG" \
-    '{decision: $decision, reason: $reason, systemMessage: $msg}'