claude-flow-novice 2.10.5 → 2.10.7

package/claude-assets/skills/cfn-ace-system/invoke-context-inject.sh

@@ -0,0 +1,361 @@
+#!/usr/bin/env bash
+
+##############################################################################
+# ACE Context Injection - Phase 3.3
+# Unified positive + negative context merging with relevance scoring
+#
+# Usage:
+#   ./invoke-context-inject.sh --task-description "task" [OPTIONS]
+#
+# Arguments:
+#   --task-description Task description for context relevance (required)
+#   --task-tags        Comma-separated tags (optional)
+#   --domain           Domain filter (optional)
+#   --enable-ace       Enable ACE context (default: true, A/B testing)
+#   --positive-limit   Max positive context bullets (default: 10)
+#   --negative-limit   Max negative context bullets (default: 5)
+#   --min-relevance    Minimum relevance score (default: 0.3)
+#   --output           Output file path (default: stdout)
+##############################################################################
+
+set -euo pipefail
+
+# Default values
+TASK_DESCRIPTION=""
+TASK_TAGS=""
+DOMAIN=""
+ENABLE_ACE=true
+POSITIVE_LIMIT=10
+NEGATIVE_LIMIT=5
+MIN_RELEVANCE=0.3
+OUTPUT=""
+
+# Parse arguments
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    --task-description)
+      TASK_DESCRIPTION="$2"
+      shift 2
+      ;;
+    --task-tags)
+      TASK_TAGS="$2"
+      shift 2
+      ;;
+    --domain)
+      DOMAIN="$2"
+      shift 2
+      ;;
+    --enable-ace)
+      ENABLE_ACE="$2"
+      shift 2
+      ;;
+    --positive-limit)
+      POSITIVE_LIMIT="$2"
+      shift 2
+      ;;
+    --negative-limit)
+      NEGATIVE_LIMIT="$2"
+      shift 2
+      ;;
+    --min-relevance)
+      MIN_RELEVANCE="$2"
+      shift 2
+      ;;
+    --output)
+      OUTPUT="$2"
+      shift 2
+      ;;
+    *)
+      echo "Unknown option: $1" >&2
+      exit 1
+      ;;
+  esac
+done
+
+# Validation
+if [ -z "$TASK_DESCRIPTION" ]; then
+  echo "Error: --task-description is required" >&2
+  echo "Usage: $0 --task-description 'task' [OPTIONS]" >&2
+  exit 1
+fi
+
+# Get project root
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PROJECT_ROOT="$(cd "$SCRIPT_DIR/../../.." && pwd)"
+
+# A/B Testing Mode - Return empty context if disabled
+if [ "$ENABLE_ACE" != "true" ]; then
+  # Log A/B test mode to Redis
+  if command -v redis-cli &> /dev/null; then
+    redis-cli HINCRBY "ace:ab_test:control_group" "invocations" 1 > /dev/null 2>&1 || true
+  fi
+
+  echo "### ACE System Context (A/B Test - Control Group)"
+  echo ""
+  echo "_ACE context disabled for A/B testing comparison._"
+  exit 0
+fi
+
+# Log A/B test mode (enabled)
+if command -v redis-cli &> /dev/null; then
+  redis-cli HINCRBY "ace:ab_test:treatment_group" "invocations" 1 > /dev/null 2>&1 || true
+fi
+
+##############################################################################
+# Relevance Scoring Function
+##############################################################################
+calculate_relevance_score() {
+  local context_tags="$1"
+  local task_tags="$2"
+
+  # No tags = domain-level relevance (0.3)
+  if [ -z "$context_tags" ] || [ "$context_tags" == "null" ]; then
+    echo "0.3"
+    return
+  fi
+
+  if [ -z "$task_tags" ]; then
+    echo "0.3"
+    return
+  fi
+
+  # Convert to arrays
+  IFS=',' read -ra CONTEXT_TAG_ARRAY <<< "$context_tags"
+  IFS=',' read -ra TASK_TAG_ARRAY <<< "$task_tags"
+
+  local max_score=0.0
+
+  # Check for tag matches
+  for ctx_tag in "${CONTEXT_TAG_ARRAY[@]}"; do
+    ctx_tag=$(echo "$ctx_tag" | xargs | tr '[:upper:]' '[:lower:]')
+
+    for task_tag in "${TASK_TAG_ARRAY[@]}"; do
+      task_tag=$(echo "$task_tag" | xargs | tr '[:upper:]' '[:lower:]')
+
+      # Exact match = 1.0
+      if [ "$ctx_tag" == "$task_tag" ]; then
+        echo "1.0"
+        return
+      fi
+
+      # Partial match = 0.6
+      if [[ "$ctx_tag" == *"$task_tag"* ]] || [[ "$task_tag" == *"$ctx_tag"* ]]; then
+        if (( $(echo "$max_score < 0.6" | bc -l) )); then
+          max_score=0.6
+        fi
+      fi
+    done
+  done
+
+  # Return max score found, or domain match (0.3)
+  if (( $(echo "$max_score > 0.0" | bc -l) )); then
+    echo "$max_score"
+  else
+    echo "0.3"
+  fi
+}
+
+##############################################################################
+# Adaptive Limit Calculation
+##############################################################################
+calculate_adaptive_limit() {
+  local relevance="$1"
+  local max_limit="$2"
+
+  # High relevance (≥0.8): Full limit
+  if (( $(echo "$relevance >= 0.8" | bc -l) )); then
+    echo "$max_limit"
+  # Medium relevance (0.5-0.8): Half limit
+  elif (( $(echo "$relevance >= 0.5" | bc -l) )); then
+    echo $((max_limit / 2))
+  # Low relevance (<0.5): Quarter limit (minimum 3)
+  else
+    local quarter=$((max_limit / 4))
+    if [ $quarter -lt 3 ]; then
+      echo "3"
+    else
+      echo "$quarter"
+    fi
+  fi
+}
+
+##############################################################################
+# Retrieve Positive Context
+##############################################################################
+retrieve_positive_context() {
+  local task_desc="$1"
+  local limit="$2"
+
+  local query_script="${SCRIPT_DIR}/query-contexts.sh"
+
+  if [ ! -f "$query_script" ]; then
+    echo "[]"
+    return
+  fi
+
+  local results
+  results=$("$query_script" "$task_desc" --limit "$limit" --format json 2>/dev/null || echo '{"results":{"contexts":[]}}')
+
+  echo "$results" | jq -r '.results.contexts[]' 2>/dev/null || echo "[]"
+}
+
+##############################################################################
+# Retrieve Negative Context
+##############################################################################
+retrieve_negative_context() {
+  local limit="$1"
+  local domain_arg=""
+  local tags_arg=""
+
+  if [ -n "$DOMAIN" ]; then
+    domain_arg="--domain $DOMAIN"
+  fi
+
+  if [ -n "$TASK_TAGS" ]; then
+    tags_arg="--task-tags $TASK_TAGS"
+  fi
+
+  local format_script="${SCRIPT_DIR}/format-negative-context.sh"
+
+  if [ ! -f "$format_script" ]; then
+    echo ""
+    return
+  fi
+
+  "$format_script" $domain_arg $tags_arg --limit "$limit" 2>/dev/null || echo ""
+}
+
+##############################################################################
+# Format Unified Context
+##############################################################################
+format_unified_context() {
+  local positive_json="$1"
+  local negative_text="$2"
+
+  echo "### ACE System Context - Adaptive Lessons"
+  echo ""
+  echo "The following lessons are derived from prior sprint iterations and system-wide knowledge."
+  echo ""
+
+  # Format positive context (strategies/patterns)
+  if [ -n "$positive_json" ] && [ "$positive_json" != "[]" ]; then
+    echo "#### Recommended Patterns & Strategies"
+    echo ""
+
+    local counter=1
+    while read -r context_line; do
+      local ctx_type=$(echo "$context_line" | jq -r '.reflection_type // "strategy"')
+      local ctx_tags=$(echo "$context_line" | jq -r '.tags // ""')
+      local ctx_confidence=$(echo "$context_line" | jq -r '.confidence // 0.0')
+      local ctx_lessons=$(echo "$context_line" | jq -r '.lessons_preview // ""')
+
+      # Calculate relevance
+      local relevance=$(calculate_relevance_score "$ctx_tags" "$TASK_TAGS")
+
+      # Skip if below minimum relevance
+      if (( $(echo "$relevance < $MIN_RELEVANCE" | bc -l) )); then
+        continue
+      fi
+
+      # Format based on type
+      case "$ctx_type" in
+        strategy)
+          echo "$counter. **Strategy** (relevance: $relevance, confidence: $ctx_confidence)"
+          ;;
+        pattern)
+          echo "$counter. **Pattern** (relevance: $relevance, confidence: $ctx_confidence)"
+          ;;
+        *)
+          echo "$counter. **Lesson** (relevance: $relevance, confidence: $ctx_confidence)"
+          ;;
+      esac
+
+      # Print lessons (truncated)
+      if [ -n "$ctx_lessons" ]; then
+        echo "   - $ctx_lessons"
+      fi
+
+      echo "   - Tags: $ctx_tags"
+      echo ""
+
+      ((counter++))
+    done < <(echo "$positive_json" | jq -c '.')
+  else
+    echo "#### Recommended Patterns & Strategies"
+    echo ""
+    echo "_No high-relevance patterns found for this task._"
+    echo ""
+  fi
+
+  # Format negative context (anti-patterns)
+  if [ -n "$negative_text" ]; then
+    echo "$negative_text"
+  else
+    echo "#### Anti-Patterns to Avoid"
+    echo ""
+    echo "_No anti-patterns found matching task criteria._"
+    echo ""
+  fi
+
+  # Metadata footer
+  echo "---"
+  echo "_Context generated by ACE System v3.3 (adaptive relevance scoring)_"
+}
+
+##############################################################################
+# Main Execution
+##############################################################################
+
+# Step 1: Retrieve positive context with adaptive limit
+POSITIVE_RESULTS=$(retrieve_positive_context "$TASK_DESCRIPTION" "$POSITIVE_LIMIT")
+
+# Step 2: Calculate average relevance for adaptive limits
+TOTAL_RELEVANCE=0.0
+COUNT=0
+
+while read -r context_line; do
+  if [ -z "$context_line" ] || [ "$context_line" == "null" ]; then
+    continue
+  fi
+
+  ctx_tags=$(echo "$context_line" | jq -r '.tags // ""')
+  relevance=$(calculate_relevance_score "$ctx_tags" "$TASK_TAGS")
+
+  TOTAL_RELEVANCE=$(echo "$TOTAL_RELEVANCE + $relevance" | bc -l)
+  ((COUNT++))
+done < <(echo "$POSITIVE_RESULTS" | jq -c '.')
+
+if [ $COUNT -gt 0 ]; then
+  AVG_RELEVANCE=$(echo "scale=2; $TOTAL_RELEVANCE / $COUNT" | bc -l)
+else
+  AVG_RELEVANCE=0.3
+fi
+
+# Step 3: Apply adaptive limits
+ADAPTIVE_POSITIVE_LIMIT=$(calculate_adaptive_limit "$AVG_RELEVANCE" "$POSITIVE_LIMIT")
+ADAPTIVE_NEGATIVE_LIMIT=$(calculate_adaptive_limit "$AVG_RELEVANCE" "$NEGATIVE_LIMIT")
+
+# Step 4: Retrieve contexts with adaptive limits
+POSITIVE_CONTEXT=$(retrieve_positive_context "$TASK_DESCRIPTION" "$ADAPTIVE_POSITIVE_LIMIT")
+NEGATIVE_CONTEXT=$(retrieve_negative_context "$ADAPTIVE_NEGATIVE_LIMIT")
+
+# Step 5: Format unified output
+UNIFIED_CONTEXT=$(format_unified_context "$POSITIVE_CONTEXT" "$NEGATIVE_CONTEXT")
+
+# Step 6: Output results
+if [ -n "$OUTPUT" ]; then
+  echo "$UNIFIED_CONTEXT" > "$OUTPUT"
+  echo "Unified context saved to: $OUTPUT" >&2
+else
+  echo "$UNIFIED_CONTEXT"
+fi
+
+# Log analytics to Redis
+if command -v redis-cli &> /dev/null; then
+  redis-cli HINCRBY "ace:stats:context_injection" "invocations" 1 > /dev/null 2>&1 || true
+  redis-cli HSET "ace:stats:context_injection" "avg_relevance" "$AVG_RELEVANCE" > /dev/null 2>&1 || true
+  redis-cli HINCRBY "ace:stats:context_injection" "positive_bullets" "$ADAPTIVE_POSITIVE_LIMIT" > /dev/null 2>&1 || true
+  redis-cli HINCRBY "ace:stats:context_injection" "negative_bullets" "$ADAPTIVE_NEGATIVE_LIMIT" > /dev/null 2>&1 || true
+fi
+
+exit 0

package/claude-assets/skills/cfn-ace-system/invoke-context-query.sh

@@ -0,0 +1,139 @@
+#!/usr/bin/env bash
+
+##############################################################################
+# ACE Context Query Wrapper
+# Searches for similar contexts using keyword matching
+#
+# Usage:
+#   ./invoke-context-query.sh --keywords "keyword1,keyword2" [OPTIONS]
+#
+# Arguments:
+#   --keywords              Comma-separated keywords (required)
+#   --similarity-threshold  Minimum similarity (0.0-1.0, default: 0.7)
+#   --max-results          Maximum results (default: 10)
+#   --memory-path          SQLite memory path (optional)
+##############################################################################
+
+set -euo pipefail
+
+# Default values
+KEYWORDS=""
+SIMILARITY_THRESHOLD=0.7
+MAX_RESULTS=10
+MEMORY_PATH="${ACE_MEMORY_PATH:-./.artifacts/database/swarm-memory.db}"
+
+# Parse arguments
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    --keywords)
+      KEYWORDS="$2"
+      shift 2
+      ;;
+    --similarity-threshold)
+      SIMILARITY_THRESHOLD="$2"
+      shift 2
+      ;;
+    --max-results)
+      MAX_RESULTS="$2"
+      shift 2
+      ;;
+    --memory-path)
+      MEMORY_PATH="$2"
+      shift 2
+      ;;
+    *)
+      echo "Unknown option: $1"
+      exit 1
+      ;;
+  esac
+done
+
+# Validation
+if [ -z "$KEYWORDS" ]; then
+  echo "Error: --keywords is required"
+  echo "Usage: $0 --keywords 'keyword1,keyword2' [OPTIONS]"
+  exit 1
+fi
+
+# Get project root
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PROJECT_ROOT="$(cd "$SCRIPT_DIR/../../.." && pwd)"
+
+# Create Node.js runner script
+RUNNER_SCRIPT=$(mktemp --suffix=.mjs)
+trap "rm -f $RUNNER_SCRIPT" EXIT
+
+cat > "$RUNNER_SCRIPT" << EOF
+import { SQLiteMemorySystem } from '${PROJECT_ROOT}/dist/memory/sqlite-memory-system.js';
+
+const keywords = process.argv[2].split(',').map(k => k.trim().toLowerCase());
+const similarityThreshold = parseFloat(process.argv[3]);
+const maxResults = parseInt(process.argv[4]);
+const memoryPath = process.argv[5];
+
+const memorySystem = new SQLiteMemorySystem(memoryPath);
+await memorySystem.initialize();
+
+// Calculate Jaccard similarity
+function jaccardSimilarity(set1, set2) {
+  const intersection = set1.filter(x => set2.includes(x));
+  const union = [...new Set([...set1, ...set2])];
+  return intersection.length / union.length;
+}
+
+// Extract keywords from context
+function extractKeywords(context) {
+  const text = JSON.stringify(context).toLowerCase();
+  const words = text.match(/\b\w+\b/g) || [];
+  return [...new Set(words)];
+}
+
+async function queryContexts() {
+  const db = memorySystem['db'];
+  if (!db) {
+    throw new Error('Database not initialized');
+  }
+
+  // Query memory_store for reflection entries
+  // Reflections are stored as: reflection:ref-{timestamp} -> JSON
+  const rows = await db.all(
+    "SELECT key, value FROM memory_store WHERE key LIKE 'reflection:%' ORDER BY created_at DESC"
+  );
+
+  if (rows.length === 0) {
+    return [];
+  }
+
+  const results = rows.map(row => {
+    // Parse the reflection JSON from BLOB/TEXT value
+    const reflection = JSON.parse(row.value);
+
+    // Extract context and keywords
+    const context = reflection.context || {};
+    const contextKeywords = extractKeywords(context);
+    const similarity = jaccardSimilarity(keywords, contextKeywords);
+
+    return {
+      id: reflection.id,
+      timestamp: reflection.timestamp,
+      complexity: reflection.complexity || 0,
+      similarity,
+      context: reflection.context,
+      insights: reflection.insights || []
+    };
+  });
+
+  // Filter by similarity threshold and sort
+  return results
+    .filter(r => r.similarity >= similarityThreshold)
+    .sort((a, b) => b.similarity - a.similarity)
+    .slice(0, maxResults);
+}
+
+const results = await queryContexts();
+console.log(JSON.stringify(results, null, 2));
+EOF
+
+# Execute query
+cd "$PROJECT_ROOT"
+node "$RUNNER_SCRIPT" "$KEYWORDS" "$SIMILARITY_THRESHOLD" "$MAX_RESULTS" "$MEMORY_PATH"