cdp-edge 1.0.0

package/extracted-skill/tracking-events-generator/agents/performance-optimization-agent.md

@@ -0,0 +1,1442 @@
+---
+name: performance-optimization-agent
+description: Performance Optimization Enterprise Agent - Caching, Query Optimization, Latency Profiling for Cloudflare Workers + D1
+type: agent
+persona: Performance Engineer specializing in edge computing optimization, distributed caching strategies, and database query optimization
+version: "1.0.0"
+---
+
+# Performance Optimization Enterprise Agent
+
+> **ESCOPO DESTE AGENTE:** Otimização ativa de performance — caching, queries, batching.
+> Para monitoramento passivo (métricas, alertas, dashboards), ver: **performance-agent.md**
+>
+> | Este agente faz | performance-agent faz |
+> |---|---|
+> | Estratégias de caching L1/L2/L3 | Medir latência e throughput |
+> | Otimização de queries D1 e indexação | Alertas de degradação de SLA |
+> | Batch processing e pipeline tuning | Health checks e dashboards |
+> | Profiling de Workers e CPU optimization | Relatórios de performance |
+
+## 🚀 Visão Geral
+
+Agente especializado em otimização de performance para o sistema CDP Edge (Cloudflare Workers + D1 + Queue). Implementa estratégias de caching multi-camada, otimização de queries, processamento em lote e monitoramento de latência em tempo real.
+
+---
+
+## 📊 Arquitetura de Performance
+
+### 3-Camadas de Caching
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│                    L1: Memory Cache                         │
+│            (In-Memory, Request Lifetime)                     │
+│            - TTL: Request duration                           │
+│            - Hit Rate: ~85%                                   │
+│            - Use: Session data, config lookup                │
+└─────────────────────────────────────────────────────────────┘
+                            ↓
+┌─────────────────────────────────────────────────────────────┐
+│                    L2: KV Cache                              │
+│            (Global, Distributed Edge)                       │
+│            - TTL: 5-60 minutes                               │
+│            - Hit Rate: ~12%                                   │
+│            - Use: Attribution data, channel metrics          │
+└─────────────────────────────────────────────────────────────┘
+                            ↓
+┌─────────────────────────────────────────────────────────────┐
+│                    L3: D1 Database                           │
+│            (SQLite at Edge, Persistent)                     │
+│            - TTL: Permanent                                  │
+│            - Use: User journeys, audit logs, security data   │
+└─────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## 🎯 Métricas de Performance
+
+### KPIs Monitorados
+
+| Métrica | Target | Alert |
+|---------|--------|-------|
+| **Cache Hit Rate (L1+L2)** | ≥ 95% | < 85% |
+| **P95 Latency (Tracking)** | < 100ms | > 200ms |
+| **P95 Latency (Attribution)** | < 500ms | > 1000ms |
+| **Query Time (D1)** | < 50ms | > 100ms |
+| **Queue Processing Time** | < 500ms | > 2000ms |
+| **Memory Usage (Worker)** | < 128MB | > 128MB |
+| **CPU Time (Worker)** | < 50ms | > 150ms |
+
+---
+
+## 💾 Caching Multi-Layer
+
+### L1: Memory Cache (Request-Scoped)
+
+```typescript
+/**
+ * L1 Memory Cache - In-request cache with automatic expiration
+ */
+class L1Cache {
+  constructor() {
+    this.cache = new Map();
+    this.stats = {
+      hits: 0,
+      misses: 0,
+      sets: 0,
+      deletes: 0
+    };
+  }
+
+  /**
+   * Get value from L1 cache
+   * @param {string} key - Cache key
+   * @returns {any|undefined} Cached value or undefined if not found/expired
+   */
+  get(key) {
+    const entry = this.cache.get(key);
+    if (!entry) {
+      this.stats.misses++;
+      return undefined;
+    }
+
+    // Check expiration
+    if (entry.expiry < Date.now()) {
+      this.cache.delete(key);
+      this.stats.misses++;
+      return undefined;
+    }
+
+    this.stats.hits++;
+    return entry.value;
+  }
+
+  /**
+   * Set value in L1 cache with optional TTL (default: request lifetime)
+   * @param {string} key - Cache key
+   * @param {any} value - Value to cache
+   * @param {number} ttlMs - TTL in milliseconds (optional)
+   */
+  set(key, value, ttlMs = null) {
+    const expiry = ttlMs ? Date.now() + ttlMs : Infinity;
+    this.cache.set(key, { value, expiry, createdAt: Date.now() });
+    this.stats.sets++;
+  }
+
+  /**
+   * Delete value from L1 cache
+   * @param {string} key - Cache key
+   */
+  delete(key) {
+    this.cache.delete(key);
+    this.stats.deletes++;
+  }
+
+  /**
+   * Clear all L1 cache entries
+   */
+  clear() {
+    this.cache.clear();
+  }
+
+  /**
+   * Get cache statistics
+   * @returns {object} Cache statistics
+   */
+  getStats() {
+    const totalRequests = this.stats.hits + this.stats.misses;
+    const hitRate = totalRequests > 0 ? (this.stats.hits / totalRequests) * 100 : 0;
+
+    return {
+      ...this.stats,
+      hitRate: hitRate.toFixed(2) + '%',
+      cacheSize: this.cache.size
+    };
+  }
+}
+
+// Global L1 cache instance (request-scoped)
+let l1Cache = new L1Cache();
+```
+
+### L2: KV Cache (Global Edge Cache)
+
+```typescript
+/**
+ * L2 KV Cache - Global distributed cache with automatic invalidation
+ */
+class L2Cache {
+  constructor(env) {
+    this.kv = env.GEO_CACHE; // Cloudflare KV namespace
+    this.stats = {
+      hits: 0,
+      misses: 0,
+      sets: 0,
+      deletes: 0
+    };
+
+    // Cache TTLs (configurable)
+    this.ttls = {
+      attribution: 3600,        // 1 hour
+      channelMetrics: 1800,     // 30 minutes
+      userJourneys: 300,        // 5 minutes
+      securityConfig: 900,      // 15 minutes
+      lookupData: 600,          // 10 minutes
+      default: 600              // 10 minutes
+    };
+  }
+
+  /**
+   * Get value from L2 cache
+   * @param {string} key - Cache key
+   * @param {string} cacheType - Type of cache (determines TTL)
+   * @returns {Promise<any|null>} Cached value or null if not found
+   */
+  async get(key, cacheType = 'default') {
+    try {
+      const cached = await this.kv.get(key, { type: 'json' });
+
+      if (cached === null) {
+        this.stats.misses++;
+        return null;
+      }
+
+      this.stats.hits++;
+      return cached;
+    } catch (error) {
+      console.error(`[L2 Cache] Error getting ${key}:`, error);
+      this.stats.misses++;
+      return null;
+    }
+  }
+
+  /**
+   * Set value in L2 cache with type-based TTL
+   * @param {string} key - Cache key
+   * @param {any} value - Value to cache
+   * @param {string} cacheType - Type of cache (determines TTL)
+   * @param {number} ttlMs - Custom TTL override (optional)
+   * @returns {Promise<boolean>} Success status
+   */
+  async set(key, value, cacheType = 'default', ttlMs = null) {
+    try {
+      const ttl = ttlMs || (this.ttls[cacheType] || this.ttls.default);
+
+      await this.kv.put(key, JSON.stringify(value), {
+        expirationTtl: ttl
+      });
+
+      this.stats.sets++;
+      return true;
+    } catch (error) {
+      console.error(`[L2 Cache] Error setting ${key}:`, error);
+      return false;
+    }
+  }
+
+  /**
+   * Delete value from L2 cache
+   * @param {string} key - Cache key
+   * @returns {Promise<boolean>} Success status
+   */
+  async delete(key) {
+    try {
+      await this.kv.delete(key);
+      this.stats.deletes++;
+      return true;
+    } catch (error) {
+      console.error(`[L2 Cache] Error deleting ${key}:`, error);
+      return false;
+    }
+  }
+
+  /**
+   * Delete multiple keys from L2 cache (batch invalidation)
+   * @param {string[]} keys - Array of cache keys
+   * @returns {Promise<number>} Number of successfully deleted keys
+   */
+  async deleteMany(keys) {
+    let deleted = 0;
+    for (const key of keys) {
+      if (await this.delete(key)) {
+        deleted++;
+      }
+    }
+    return deleted;
+  }
+
+  /**
+   * List keys by prefix (for cache invalidation)
+   * @param {string} prefix - Key prefix
+   * @param {number} limit - Maximum keys to return
+   * @returns {Promise<string[]>} Array of keys
+   */
+  async listKeys(prefix, limit = 100) {
+    try {
+      const list = await this.kv.list({ prefix, limit });
+      return list.keys.map(k => k.name);
+    } catch (error) {
+      console.error(`[L2 Cache] Error listing keys with prefix ${prefix}:`, error);
+      return [];
+    }
+  }
+
+  /**
+   * Invalidate cache by prefix (bulk invalidation)
+   * @param {string} prefix - Key prefix to invalidate
+   * @returns {Promise<number>} Number of invalidated keys
+   */
+  async invalidateByPrefix(prefix) {
+    const keys = await this.listKeys(prefix, 1000);
+    return await this.deleteMany(keys);
+  }
+
+  /**
+   * Get cache statistics
+   * @returns {object} Cache statistics
+   */
+  getStats() {
+    const totalRequests = this.stats.hits + this.stats.misses;
+    const hitRate = totalRequests > 0 ? (this.stats.hits / totalRequests) * 100 : 0;
+
+    return {
+      ...this.stats,
+      hitRate: hitRate.toFixed(2) + '%'
+    };
+  }
+}
+
+// L2 cache instance (environment-scoped)
+let l2Cache = null;
+```
+
+### Cache Strategy Implementation
+
+```typescript
+/**
+ * Multi-Layer Cache Manager
+ * Orchestrates L1 and L2 caches with fallback logic
+ */
+class CacheManager {
+  constructor(env) {
+    this.l1 = new L1Cache();
+    this.l2 = new L2Cache(env);
+  }
+
+  /**
+   * Get value from cache (L1 → L2 → D1 fallback)
+   * @param {string} key - Cache key
+   * @param {string} cacheType - Type of cache
+   * @param {Function} fallback - Fallback function to fetch from D1
+   * @returns {Promise<any>} Cached or fetched value
+   */
+  async get(key, cacheType = 'default', fallback = null) {
+    // Try L1 cache first (fastest)
+    const l1Value = this.l1.get(key);
+    if (l1Value !== undefined) {
+      return l1Value;
+    }
+
+    // Try L2 cache (distributed)
+    const l2Value = await this.l2.get(key, cacheType);
+    if (l2Value !== null) {
+      // Promote to L1 cache
+      this.l1.set(key, l2Value);
+      return l2Value;
+    }
+
+    // Fallback to source (D1)
+    if (fallback) {
+      const value = await fallback();
+
+      // Cache in L2 and L1
+      await this.l2.set(key, value, cacheType);
+      this.l1.set(key, value);
+
+      return value;
+    }
+
+    return null;
+  }
+
+  /**
+   * Set value in cache (L1 + L2)
+   * @param {string} key - Cache key
+   * @param {any} value - Value to cache
+   * @param {string} cacheType - Type of cache
+   * @param {number} ttlMs - Custom TTL for L2
+   */
+  async set(key, value, cacheType = 'default', ttlMs = null) {
+    this.l1.set(key, value);
+    await this.l2.set(key, value, cacheType, ttlMs);
+  }
+
+  /**
+   * Delete value from cache (L1 + L2)
+   * @param {string} key - Cache key
+   */
+  async delete(key) {
+    this.l1.delete(key);
+    await this.l2.delete(key);
+  }
+
+  /**
+   * Invalidate cache by prefix (L2 only - L1 doesn't support prefix)
+   * @param {string} prefix - Key prefix
+   */
+  async invalidateByPrefix(prefix) {
+    await this.l2.invalidateByPrefix(prefix);
+  }
+
+  /**
+   * Get combined cache statistics
+   * @returns {object} Combined cache statistics
+   */
+  getStats() {
+    return {
+      l1: this.l1.getStats(),
+      l2: this.l2.getStats()
+    };
+  }
+}
+
+// Global cache manager instance
+let cacheManager = null;
+```
+
+---
+
+## 🗄️ Otimização de Queries D1
+
+### Schema de Índices Otimizados
+
+```sql
+-- User Journeys Table (Optimized for attribution queries)
+CREATE TABLE IF NOT EXISTS user_journeys (
+  id INTEGER PRIMARY KEY AUTOINCREMENT,
+  user_id TEXT NOT NULL,
+  event_id TEXT,
+  event_name TEXT NOT NULL,
+  utm_source TEXT,
+  utm_medium TEXT,
+  utm_campaign TEXT,
+  utm_term TEXT,
+  utm_content TEXT,
+  event_timestamp DATETIME NOT NULL,
+  position INTEGER,
+  conversion_id TEXT,
+  created_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+-- Compound indexes for common query patterns
+CREATE INDEX IF NOT EXISTS idx_user_journeys_user_timestamp
+  ON user_journeys(user_id, event_timestamp DESC);
+
+CREATE INDEX IF NOT EXISTS idx_user_journeys_conversion
+  ON user_journeys(conversion_id, position);
+
+CREATE INDEX IF NOT EXISTS idx_user_journeys_utm_source
+  ON user_journeys(utm_source, event_timestamp DESC);
+
+CREATE INDEX IF NOT EXISTS idx_user_journeys_event_name
+  ON user_journeys(event_name, event_timestamp DESC);
+
+-- Multi-Touch Attribution Table
+CREATE TABLE IF NOT EXISTS multi_touch_attribution (
+  conversion_id TEXT NOT NULL,
+  attribution_model TEXT NOT NULL,
+  touchpoint_index INTEGER NOT NULL,
+  utm_source TEXT,
+  credit_percentage REAL NOT NULL,
+  role TEXT,
+  created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+  PRIMARY KEY (conversion_id, attribution_model, touchpoint_index)
+);
+
+-- Index for attribution queries
+CREATE INDEX IF NOT EXISTS idx_attribution_model_conversion
+  ON multi_touch_attribution(attribution_model, conversion_id);
+
+CREATE INDEX IF NOT EXISTS idx_attribution_source
+  ON multi_touch_attribution(utm_source, attribution_model);
+
+-- Channel Performance Table (Materialized for fast aggregation)
+CREATE TABLE IF NOT EXISTS channel_performance (
+  utm_source TEXT NOT NULL,
+  attribution_model TEXT NOT NULL,
+  total_attribution REAL NOT NULL,
+  total_conversions INTEGER NOT NULL,
+  total_value REAL NOT NULL,
+  avg_attribution_per_conversion REAL NOT NULL,
+  last_updated DATETIME NOT NULL,
+  PRIMARY KEY (utm_source, attribution_model)
+);
+
+-- Index for performance queries
+CREATE INDEX IF NOT EXISTS idx_channel_performance_model
+  ON channel_performance(attribution_model, total_attribution DESC);
+
+-- Index for timestamp-based queries
+CREATE INDEX IF NOT EXISTS idx_channel_performance_updated
+  ON channel_performance(last_updated DESC);
+```
+
+### Query Optimization Patterns
+
+```typescript
+/**
+ * Optimized Query Builder for D1
+ */
+class QueryOptimizer {
+  constructor(db) {
+    this.db = db;
+    this.queryCache = new Map();
+  }
+
+  /**
+   * Execute query with automatic index usage and caching
+   * @param {string} query - SQL query with parameter placeholders
+   * @param {any[]} params - Query parameters
+   * @param {string} cacheKey - Optional cache key
+   * @returns {Promise<any>} Query result
+   */
+  async execute(query, params = [], cacheKey = null) {
+    const startTime = performance.now();
+
+    try {
+      // Check query cache if cache key provided
+      if (cacheKey && this.queryCache.has(cacheKey)) {
+        const cached = this.queryCache.get(cacheKey);
+        if (Date.now() - cached.timestamp < 60000) { // 1 min cache
+          return cached.result;
+        }
+      }
+
+      // Prepare and execute query
+      const statement = this.db.prepare(query);
+      const result = await statement.bind(...params).all();
+
+      // Cache result if cache key provided
+      if (cacheKey) {
+        this.queryCache.set(cacheKey, {
+          result: result,
+          timestamp: Date.now()
+        });
+      }
+
+      const queryTime = performance.now() - startTime;
+
+      // Log slow queries
+      if (queryTime > 100) {
+        console.warn(`[Slow Query] ${queryTime.toFixed(2)}ms - ${query.substring(0, 100)}...`);
+      }
+
+      return result;
+    } catch (error) {
+      const queryTime = performance.now() - startTime;
+      console.error(`[Query Error] ${queryTime.toFixed(2)}ms - ${error.message}`);
+      throw error;
+    }
+  }
+
+  /**
+   * Get user journey with optimized index usage
+   * @param {string} userId - User ID
+   * @param {string} conversionId - Optional conversion ID filter
+   * @returns {Promise<Array>} User journey events
+   */
+  async getUserJourney(userId, conversionId = null) {
+    const cacheKey = `journey:${userId}:${conversionId || 'all'}`;
+
+    const query = conversionId
+      ? `
+        SELECT * FROM user_journeys
+        WHERE user_id = ? AND conversion_id = ?
+        ORDER BY event_timestamp ASC
+      `
+      : `
+        SELECT * FROM user_journeys
+        WHERE user_id = ?
+        ORDER BY event_timestamp DESC
+        LIMIT 100
+      `;
+
+    const params = conversionId ? [userId, conversionId] : [userId];
+    return await this.execute(query, params, cacheKey);
+  }
+
+  /**
+   * Get attribution data with optimized joins
+   * @param {string} conversionId - Conversion ID
+   * @param {string} attributionModel - Attribution model
+   * @returns {Promise<Array>} Attribution data
+   */
+  async getAttributionData(conversionId, attributionModel) {
+    const cacheKey = `attribution:${conversionId}:${attributionModel}`;
+
+    const query = `
+      SELECT
+        a.*,
+        j.utm_medium,
+        j.utm_campaign,
+        j.event_name,
+        j.event_timestamp
+      FROM multi_touch_attribution a
+      INNER JOIN user_journeys j ON a.conversion_id = j.conversion_id
+      WHERE a.conversion_id = ? AND a.attribution_model = ?
+      ORDER BY a.touchpoint_index ASC
+    `;
+
+    return await this.execute(query, [conversionId, attributionModel], cacheKey);
+  }
+
+  /**
+   * Get channel performance with pre-aggregated data
+   * @param {string} attributionModel - Attribution model
+   * @param {number} limit - Limit results
+   * @returns {Promise<Array>} Channel performance data
+   */
+  async getChannelPerformance(attributionModel, limit = 50) {
+    const cacheKey = `performance:${attributionModel}:${limit}`;
+
+    const query = `
+      SELECT * FROM channel_performance
+      WHERE attribution_model = ?
+      ORDER BY total_attribution DESC
+      LIMIT ?
+    `;
+
+    return await this.execute(query, [attributionModel, limit], cacheKey);
+  }
+
+  /**
+   * Batch insert for high-throughput operations
+   * @param {string} table - Table name
+   * @param {string[]} columns - Column names
+   * @param {any[][]} values - Array of value arrays
+   * @returns {Promise<number>} Number of inserted rows
+   */
+  async batchInsert(table, columns, values) {
+    const startTime = performance.now();
+
+    try {
+      const placeholders = columns.map(() => '?').join(',');
+      const query = `INSERT INTO ${table} (${columns.join(',')}) VALUES (${placeholders})`;
+
+      let inserted = 0;
+      for (const row of values) {
+        await this.db.prepare(query).bind(...row).run();
+        inserted++;
+      }
+
+      const queryTime = performance.now() - startTime;
+      console.log(`[Batch Insert] ${inserted} rows in ${queryTime.toFixed(2)}ms`);
+
+      return inserted;
+    } catch (error) {
+      const queryTime = performance.now() - startTime;
+      console.error(`[Batch Insert Error] ${queryTime.toFixed(2)}ms - ${error.message}`);
+      throw error;
+    }
+  }
+
+  /**
+   * Invalidate query cache by pattern
+   * @param {string} pattern - Cache key pattern (supports wildcards *)
+   */
+  invalidateCache(pattern) {
+    for (const key of this.queryCache.keys()) {
+      if (this.matchesPattern(key, pattern)) {
+        this.queryCache.delete(key);
+      }
+    }
+  }
+
+  /**
+   * Match cache key against pattern (supports * wildcard)
+   * @param {string} key - Cache key
+   * @param {string} pattern - Pattern with * wildcards
+   * @returns {boolean} Match result
+   */
+  matchesPattern(key, pattern) {
+    const regex = new RegExp('^' + pattern.replace(/\*/g, '.*') + '$');
+    return regex.test(key);
+  }
+}
+
+// Global query optimizer instance
+let queryOptimizer = null;
+```
+
+---
+
+## 🚀 Otimização de Processamento em Lote
+
+### Batch Processing for Attribution
+
+```typescript
+/**
+ * Batch Attribution Processor
+ * Processes multiple attributions in parallel with batching
+ */
+class BatchAttributionProcessor {
+  constructor(cacheManager, queryOptimizer) {
+    this.cache = cacheManager;
+    this.optimizer = queryOptimizer;
+    this.batchSize = 50; // Process 50 attributions at once
+    this.maxParallel = 5; // Max 5 parallel batches
+  }
+
+  /**
+   * Process multiple attributions in batches
+   * @param {Array} conversions - Array of conversion IDs
+   * @param {string} attributionModel - Attribution model
+   * @returns {Promise<Array>} Processing results
+   */
+  async processBatch(conversions, attributionModel) {
+    const results = [];
+    const batches = this.chunkArray(conversions, this.batchSize);
+
+    // Process batches in parallel with concurrency limit
+    const batchPromises = batches.map(batch =>
+      this.processSingleBatch(batch, attributionModel)
+    );
+
+    const batchResults = await Promise.all(batchPromises);
+
+    // Flatten results
+    for (const batchResult of batchResults) {
+      results.push(...batchResult);
+    }
+
+    return results;
+  }
+
+  /**
+   * Process a single batch of attributions
+   * @param {Array} conversions - Array of conversion IDs
+   * @param {string} attributionModel - Attribution model
+   * @returns {Promise<Array>} Processing results
+   */
+  async processSingleBatch(conversions, attributionModel) {
+    const results = [];
+
+    for (const conversionId of conversions) {
+      try {
+        // Check cache first
+        const cacheKey = `attribution:${conversionId}:${attributionModel}`;
+        const cached = await this.cache.get(cacheKey, 'attribution');
+
+        if (cached) {
+          results.push({ conversionId, attributionModel, cached, fromCache: true });
+          continue;
+        }
+
+        // Fetch user journey
+        const journey = await this.optimizer.getUserJourney(conversionId);
+
+        if (journey.length === 0) {
+          results.push({ conversionId, attributionModel, error: 'No journey found' });
+          continue;
+        }
+
+        // Calculate attribution
+        const attribution = await this.calculateAttribution(journey, attributionModel);
+
+        // Cache result
+        await this.cache.set(cacheKey, attribution, 'attribution');
+
+        results.push({ conversionId, attributionModel, attribution, fromCache: false });
+      } catch (error) {
+        console.error(`Error processing ${conversionId}:`, error);
+        results.push({ conversionId, attributionModel, error: error.message });
+      }
+    }
+
+    return results;
+  }
+
+  /**
+   * Calculate attribution using specified model
+   * @param {Array} journey - User journey events
+   * @param {string} model - Attribution model
+   * @returns {Array} Attribution result
+   */
+  async calculateAttribution(journey, model) {
+    // Implementation of attribution models
+    // This would call the appropriate attribution function from attribution-agent.md
+    const attributionModels = {
+      'last_click': this.lastClickAttribution,
+      'first_click': this.firstClickAttribution,
+      'linear': this.linearAttribution,
+      'time_decay': this.timeDecayAttribution,
+      'u_shape': this.uShapeAttribution,
+      'w_shape': this.wShapeAttribution,
+      'data_driven': this.dataDrivenAttribution
+    };
+
+    const modelFn = attributionModels[model] || attributionModels['last_click'];
+    return await modelFn.call(this, journey);
+  }
+
+  // Attribution model implementations (from attribution-agent.md)
+  lastClickAttribution(touchpoints) {
+    return touchpoints.map((tp, index) => ({
+      ...tp,
+      credit_percentage: index === touchpoints.length - 1 ? 100 : 0,
+      is_last_click: index === touchpoints.length - 1
+    }));
+  }
+
+  firstClickAttribution(touchpoints) {
+    return touchpoints.map((tp, index) => ({
+      ...tp,
+      credit_percentage: index === 0 ? 100 : 0,
+      is_first_click: index === 0
+    }));
+  }
+
+  linearAttribution(touchpoints) {
+    const credit = 100 / touchpoints.length;
+    return touchpoints.map(tp => ({
+      ...tp,
+      credit_percentage: credit.toFixed(2)
+    }));
+  }
+
+  async timeDecayAttribution(touchpoints, decayFactor = 0.9) {
+    const now = Date.now();
+    const oneDayMs = 24 * 60 * 60 * 1000;
+
+    const touchpointsWithDays = touchpoints.map(tp => {
+      const daysSinceTouch = (now - tp.event_timestamp) / oneDayMs;
+      const decayScore = Math.pow(decayFactor, daysSinceTouch);
+      return { ...tp, days_since: daysSinceTouch, decay_score: decayScore };
+    });
+
+    const totalScore = touchpointsWithDays.reduce((sum, tp) => sum + tp.decay_score, 0);
+
+    return touchpointsWithDays.map(tp => ({
+      ...tp,
+      credit_percentage: ((tp.decay_score / totalScore) * 100).toFixed(2)
+    }));
+  }
+
+  uShapeAttribution(touchpoints) {
+    if (touchpoints.length === 0) return [];
+
+    const firstTouchWeight = 0.4;
+    const lastTouchWeight = 0.4;
+    const middleWeight = touchpoints.length > 2 ? 0.2 / (touchpoints.length - 2) : 0;
+
+    return touchpoints.map((tp, index) => {
+      if (index === 0) {
+        return { ...tp, credit_percentage: 40, role: 'first' };
+      } else if (index === touchpoints.length - 1) {
+        return { ...tp, credit_percentage: 40, role: 'last' };
+      } else {
+        return { ...tp, credit_percentage: middleWeight.toFixed(2), role: 'middle' };
+      }
+    });
+  }
+
+  wShapeAttribution(touchpoints) {
+    if (touchpoints.length === 0) return [];
+
+    const firstTouchWeight = 0.3;
+    const lastTouchWeight = 0.3;
+    const middleWeights = touchpoints.length > 2 ? 0.4 / 2 : 0; // 2 assist points
+
+    return touchpoints.map((tp, index) => {
+      const pos = index + 1;
+      const total = touchpoints.length;
+
+      if (pos === 1) {
+        return { ...tp, credit_percentage: 30, role: 'first' };
+      } else if (pos === total) {
+        return { ...tp, credit_percentage: 30, role: 'last' };
+      } else if (pos === Math.floor(total / 2)) {
+        return { ...tp, credit_percentage: 20, role: 'middle_assist_1' };
+      } else {
+        return { ...tp, credit_percentage: middleWeights.toFixed(2), role: 'middle_assist_2' };
+      }
+    });
+  }
+
+  async dataDrivenAttribution(touchpoints, userJourneyHistory) {
+    // Simplified data-driven attribution
+    // In production, this would analyze historical conversion data
+
+    const channelWeights = await this.calculateChannelWeights(touchpoints, userJourneyHistory);
+    const positionWeights = await this.calculatePositionWeights(touchpoints, userJourneyHistory);
+
+    return touchpoints.map((tp, index) => {
+      const baseScore = 100 / touchpoints.length;
+      const channelWeight = channelWeights[tp.utm_source] || 1;
+      const positionWeight = positionWeights[index] || 1;
+
+      const score = baseScore * channelWeight * positionWeight;
+
+      return {
+        ...tp,
+        credit_percentage: score.toFixed(2),
+        channel_weight: channelWeight,
+        position_weight: positionWeight
+      };
+    });
+  }
+
+  async calculateChannelWeights(touchpoints, history) {
+    // Simplified - in production, analyze historical data
+    const channels = {};
+    touchpoints.forEach(tp => {
+      channels[tp.utm_source] = (channels[tp.utm_source] || 0) + 1;
+    });
+
+    const total = Object.values(channels).reduce((a, b) => a + b, 0);
+    const weights = {};
+    Object.entries(channels).forEach(([channel, count]) => {
+      weights[channel] = (count / total) * touchpoints.length;
+    });
+
+    return weights;
+  }
+
+  async calculatePositionWeights(touchpoints, history) {
+    // Simplified - in production, analyze historical position conversion rates
+    const weights = {};
+    for (let i = 0; i < touchpoints.length; i++) {
+      weights[i] = 1;
+    }
+    return weights;
+  }
+
+  /**
+   * Chunk array into smaller arrays
+   * @param {Array} array - Array to chunk
+   * @param {number} size - Chunk size
+   * @returns {Array} Array of chunks
+   */
+  chunkArray(array, size) {
+    const chunks = [];
+    for (let i = 0; i < array.length; i += size) {
+      chunks.push(array.slice(i, i + size));
+    }
+    return chunks;
+  }
+}
+
+// Global batch processor instance
+let batchProcessor = null;
+```
+
+---
+
+## 📊 Profileamento de Latência
+
+### Performance Monitoring
+
+```typescript
+/**
+ * Latency Profiler
+ * Monitors and tracks performance metrics
+ */
+class LatencyProfiler {
+  constructor() {
+    this.metrics = {
+      requests: [],
+      queries: [],
+      cacheOps: [],
+      externalAPIs: []
+    };
+
+    this.thresholds = {
+      request: { warning: 100, error: 200 },
+      query: { warning: 50, error: 100 },
+      cache: { warning: 10, error: 20 },
+      external: { warning: 500, error: 2000 }
+    };
+  }
+
+  /**
+   * Start profiling a request
+   * @param {string} requestId - Request ID
+   * @param {string} endpoint - Endpoint being accessed
+   */
+  startRequest(requestId, endpoint) {
+    this.metrics.requests.push({
+      id: requestId,
+      endpoint: endpoint,
+      startTime: performance.now(),
+      stages: []
+    });
+  }
+
+  /**
+   * End profiling a request
+   * @param {string} requestId - Request ID
+   * @param {object} metadata - Additional metadata
+   */
+  endRequest(requestId, metadata = {}) {
+    const request = this.metrics.requests.find(r => r.id === requestId);
+    if (!request) return;
+
+    request.endTime = performance.now();
+    request.duration = request.endTime - request.startTime;
+    request.metadata = metadata;
+
+    // Log if over threshold
+    if (request.duration > this.thresholds.request.error) {
+      console.error(`[Latency] CRITICAL: ${request.duration.toFixed(2)}ms - ${request.endpoint}`);
+    } else if (request.duration > this.thresholds.request.warning) {
+      console.warn(`[Latency] WARNING: ${request.duration.toFixed(2)}ms - ${request.endpoint}`);
+    }
+  }
+
+  /**
+   * Profile a D1 query
+   * @param {string} requestId - Request ID
+   * @param {string} query - SQL query
+   * @returns {Function} Function to call when query completes
+   */
+  profileQuery(requestId, query) {
+    const startTime = performance.now();
+
+    return () => {
+      const duration = performance.now() - startTime;
+
+      this.metrics.queries.push({
+        requestId: requestId,
+        query: query.substring(0, 100),
+        duration: duration,
+        timestamp: Date.now()
+      });
+
+      if (duration > this.thresholds.query.error) {
+        console.error(`[Query Latency] CRITICAL: ${duration.toFixed(2)}ms - ${query.substring(0, 50)}...`);
+      } else if (duration > this.thresholds.query.warning) {
+        console.warn(`[Query Latency] WARNING: ${duration.toFixed(2)}ms - ${query.substring(0, 50)}...`);
+      }
+    };
+  }
+
+  /**
+   * Profile a cache operation
+   * @param {string} requestId - Request ID
+   * @param {string} operation - Cache operation (get, set, delete)
+   * @param {string} layer - Cache layer (L1, L2)
+   * @returns {Function} Function to call when operation completes
+   */
+  profileCache(requestId, operation, layer) {
+    const startTime = performance.now();
+
+    return () => {
+      const duration = performance.now() - startTime;
+
+      this.metrics.cacheOps.push({
+        requestId: requestId,
+        operation: operation,
+        layer: layer,
+        duration: duration,
+        timestamp: Date.now()
+      });
+
+      if (duration > this.thresholds.cache.error) {
+        console.error(`[Cache Latency] CRITICAL: ${layer} ${operation} - ${duration.toFixed(2)}ms`);
+      }
+    };
+  }
+
+  /**
+   * Profile an external API call
+   * @param {string} requestId - Request ID
+   * @param {string} api - API name
+   * @returns {Function} Function to call when API call completes
+   */
+  profileExternalAPI(requestId, api) {
+    const startTime = performance.now();
+
+    return () => {
+      const duration = performance.now() - startTime;
+
+      this.metrics.externalAPIs.push({
+        requestId: requestId,
+        api: api,
+        duration: duration,
+        timestamp: Date.now()
+      });
+
+      if (duration > this.thresholds.external.error) {
+        console.error(`[External API Latency] CRITICAL: ${api} - ${duration.toFixed(2)}ms`);
+      } else if (duration > this.thresholds.external.warning) {
+        console.warn(`[External API Latency] WARNING: ${api} - ${duration.toFixed(2)}ms`);
+      }
+    };
+  }
+
+  /**
+   * Get performance summary
+   * @returns {object} Performance summary
+   */
+  getSummary() {
+    const summary = {};
+
+    // Request metrics
+    const completedRequests = this.metrics.requests.filter(r => r.duration);
+    summary.requests = {
+      total: completedRequests.length,
+      avg: this.average(completedRequests.map(r => r.duration)),
+      p50: this.percentile(completedRequests.map(r => r.duration), 50),
+      p95: this.percentile(completedRequests.map(r => r.duration), 95),
+      p99: this.percentile(completedRequests.map(r => r.duration), 99),
+      max: Math.max(...completedRequests.map(r => r.duration), 0)
+    };
+
+    // Query metrics
+    summary.queries = {
+      total: this.metrics.queries.length,
+      avg: this.average(this.metrics.queries.map(q => q.duration)),
+      p95: this.percentile(this.metrics.queries.map(q => q.duration), 95),
+      max: Math.max(...this.metrics.queries.map(q => q.duration), 0)
+    };
+
+    // Cache metrics
+    const l1Ops = this.metrics.cacheOps.filter(o => o.layer === 'L1');
+    const l2Ops = this.metrics.cacheOps.filter(o => o.layer === 'L2');
+
+    summary.cache = {
+      l1: {
+        total: l1Ops.length,
+        avg: this.average(l1Ops.map(o => o.duration)),
+        hitRate: this.calculateHitRate(l1Ops)
+      },
+      l2: {
+        total: l2Ops.length,
+        avg: this.average(l2Ops.map(o => o.duration)),
+        hitRate: this.calculateHitRate(l2Ops)
+      }
+    };
+
+    // External API metrics
+    summary.external = {
+      total: this.metrics.externalAPIs.length,
+      avg: this.average(this.metrics.externalAPIs.map(a => a.duration)),
+      p95: this.percentile(this.metrics.externalAPIs.map(a => a.duration), 95),
+      byAPI: this.groupByAPI()
+    };
+
+    return summary;
+  }
+
+  /**
+   * Calculate average of array
+   * @param {number[]} arr - Array of numbers
+   * @returns {number} Average
+   */
+  average(arr) {
+    if (arr.length === 0) return 0;
+    return arr.reduce((a, b) => a + b, 0) / arr.length;
+  }
+
+  /**
+   * Calculate percentile of array
+   * @param {number[]} arr - Array of numbers
+   * @param {number} p - Percentile (0-100)
+   * @returns {number} Percentile value
+   */
+  percentile(arr, p) {
+    if (arr.length === 0) return 0;
+    const sorted = arr.sort((a, b) => a - b);
+    const index = Math.ceil((p / 100) * sorted.length) - 1;
+    return sorted[index];
+  }
+
+  /**
+   * Calculate cache hit rate
+   * @param {Array} ops - Cache operations
+   * @returns {number} Hit rate percentage
+   */
+  calculateHitRate(ops) {
+    const hits = ops.filter(o => o.operation === 'get').length;
+    const total = ops.length;
+    return total > 0 ? ((hits / total) * 100).toFixed(2) + '%' : '0%';
+  }
+
+  /**
+   * Group external API calls by API
+   * @returns {object} Grouped metrics
+   */
+  groupByAPI() {
+    const grouped = {};
+    this.metrics.externalAPIs.forEach(call => {
+      if (!grouped[call.api]) {
+        grouped[call.api] = { total: 0, durations: [] };
+      }
+      grouped[call.api].total++;
+      grouped[call.api].durations.push(call.duration);
+    });
+
+    Object.keys(grouped).forEach(api => {
+      grouped[api].avg = this.average(grouped[api].durations);
+      grouped[api].p95 = this.percentile(grouped[api].durations, 95);
+    });
+
+    return grouped;
+  }
+
+  /**
+   * Clear all metrics
+   */
+  clear() {
+    this.metrics = {
+      requests: [],
+      queries: [],
+      cacheOps: [],
+      externalAPIs: []
+    };
+  }
+}
+
+// Global latency profiler instance
+let latencyProfiler = new LatencyProfiler();
+```
+
+---
+
+## 🔧 Endpoints de Monitoramento de Performance
+
+### Worker API Endpoints
+
+```typescript
+/**
+ * Performance Monitoring Endpoints
+ */
+export default {
+  async fetch(request, env, ctx) {
+    // Initialize components
+    if (!cacheManager) cacheManager = new CacheManager(env);
+    if (!queryOptimizer) queryOptimizer = new QueryOptimizer(env.DB);
+    if (!batchProcessor) batchProcessor = new BatchAttributionProcessor(cacheManager, queryOptimizer);
+
+    const url = new URL(request.url);
+    const path = url.pathname;
+
+    // Performance monitoring endpoints
+    if (path === '/api/performance/stats') {
+      return handlePerformanceStats(env);
+    }
+
+    if (path === '/api/performance/cache-stats') {
+      return handleCacheStats(env);
+    }
+
+    if (path === '/api/performance/query-stats') {
+      return handleQueryStats(env);
+    }
+
+    if (path === '/api/performance/latency-summary') {
+      return handleLatencySummary(env);
+    }
+
+    if (path === '/api/performance/cache-invalidate') {
+      return handleCacheInvalidate(request, env);
+    }
+
+    // ... other endpoints
+
+    return new Response('Not Found', { status: 404 });
+  }
+};
+
+/**
+ * Handle /api/performance/stats - Overall performance statistics
+ */
+async function handlePerformanceStats(env) {
+  const summary = latencyProfiler.getSummary();
+
+  return new Response(JSON.stringify({
+    success: true,
+    data: {
+      performance: summary,
+      timestamp: new Date().toISOString()
+    }
+  }), {
+    headers: { 'Content-Type': 'application/json' }
+  });
+}
+
+/**
+ * Handle /api/performance/cache-stats - Cache statistics
+ */
+async function handleCacheStats(env) {
+  const stats = cacheManager ? cacheManager.getStats() : null;
+
+  return new Response(JSON.stringify({
+    success: true,
+    data: {
+      cache: stats,
+      timestamp: new Date().toISOString()
+    }
+  }), {
+    headers: { 'Content-Type': 'application/json' }
+  });
+}
+
+/**
+ * Handle /api/performance/query-stats - Query statistics
+ */
+async function handleQueryStats(env) {
+  const queryMetrics = latencyProfiler.metrics.queries;
+
+  return new Response(JSON.stringify({
+    success: true,
+    data: {
+      queries: {
+        total: queryMetrics.length,
+        avg: queryMetrics.length > 0
+          ? queryMetrics.reduce((a, b) => a + b.duration, 0) / queryMetrics.length
+          : 0,
+        slowQueries: queryMetrics.filter(q => q.duration > 100),
+        timestamp: new Date().toISOString()
+      }
+    }
+  }), {
+    headers: { 'Content-Type': 'application/json' }
+  });
+}
+
+/**
+ * Handle /api/performance/latency-summary - Latency summary
+ */
+async function handleLatencySummary(env) {
+  const summary = latencyProfiler.getSummary();
+
+  return new Response(JSON.stringify({
+    success: true,
+    data: {
+      latency: {
+        requests: summary.requests,
+        queries: summary.queries,
+        cache: summary.cache,
+        externalAPIs: summary.external
+      },
+      timestamp: new Date().toISOString()
+    }
+  }), {
+    headers: { 'Content-Type': 'application/json' }
+  });
+}
+
+/**
+ * Handle /api/performance/cache-invalidate - Cache invalidation
+ */
+async function handleCacheInvalidate(request, env) {
+  if (request.method !== 'POST') {
+    return new Response('Method Not Allowed', { status: 405 });
+  }
+
+  try {
+    const body = await request.json();
+    const { prefix, keys } = body;
+
+    if (prefix) {
+      await cacheManager.invalidateByPrefix(prefix);
+    } else if (keys && Array.isArray(keys)) {
+      for (const key of keys) {
+        await cacheManager.delete(key);
+      }
+    } else {
+      return new Response(JSON.stringify({
+        success: false,
+        error: 'Must provide either prefix or keys'
+      }), {
+        status: 400,
+        headers: { 'Content-Type': 'application/json' }
+      });
+    }
+
+    return new Response(JSON.stringify({
+      success: true,
+      message: 'Cache invalidated successfully',
+      timestamp: new Date().toISOString()
+    }), {
+      headers: { 'Content-Type': 'application/json' }
+    });
+  } catch (error) {
+    return new Response(JSON.stringify({
+      success: false,
+      error: error.message
+    }), {
+      status: 500,
+      headers: { 'Content-Type': 'application/json' }
+    });
+  }
+}
+```
+
+---
+
+## 📋 Resumo de Implementação
+
+### Componentes Criados
+
+1. **CacheManager** - Gerenciador de caching multi-camada (L1 + L2 + L3)
+2. **QueryOptimizer** - Otimizador de queries com cache e índices
+3. **BatchAttributionProcessor** - Processamento em lote de atribuições
+4. **LatencyProfiler** - Monitoramento de latência em tempo real
+
+### Endpoints de Monitoramento
+
+| Endpoint | Método | Descrição |
+|----------|--------|-----------|
+| `/api/performance/stats` | GET | Estatísticas gerais de performance |
+| `/api/performance/cache-stats` | GET | Estatísticas de cache (L1/L2) |
+| `/api/performance/query-stats` | GET | Estatísticas de queries |
+| `/api/performance/latency-summary` | GET | Resumo de latência (P50, P95, P99) |
+| `/api/performance/cache-invalidate` | POST | Invalidação de cache |
+
+### Melhorias de Performance
+
+- **Cache Hit Rate**: > 95% (L1 + L2)
+- **P95 Latency**: < 100ms (tracking), < 500ms (attribution)
+- **Query Time**: < 50ms (queries otimizadas)
+- **Memory Usage**: < 128MB (workers)
+- **CPU Time**: < 50ms (workers)
+
+---
+
+## 🔗 Integração com Outros Agentes
+
+- **attribution-agent.md**: Usa `BatchAttributionProcessor` para processar múltiplas atribuições em paralelo
+- **security-enterprise-agent.md**: Usa `CacheManager` para cache de configurações de segurança
+- **master-orchestrator.md**: Usa `LatencyProfiler` para monitorar performance de toda a pipeline
+
+---
+
+## ✅ Checklist de Implementação
+
+- [x] Implementar L1 Memory Cache (request-scoped)
+- [x] Implementar L2 KV Cache (global edge cache)
+- [x] Implementar CacheManager com fallback L1 → L2 → D1
+- [x] Criar índices otimizados para D1
+- [x] Implementar QueryOptimizer com cache de queries
+- [x] Implementar BatchAttributionProcessor para processamento em lote
+- [x] Implementar LatencyProfiler com monitoramento em tempo real
+- [x] Criar endpoints de monitoramento de performance
+- [x] Implementar invalidação de cache por prefixo
+- [x] Configurar TTLs específicos por tipo de cache
+
+---
+
+## 📚 Referências
+
+- [Cloudflare Workers Performance Best Practices](https://developers.cloudflare.com/workers/)
+- [Cloudflare KV Caching](https://developers.cloudflare.com/workers/runtime-apis/kv/)
+- [Cloudflare D1 Query Optimization](https://developers.cloudflare.com/d1/)
+- [SQLite Index Optimization](https://www.sqlite.org/queryplanner.html)
+
+---
+
+*Performance Optimization Enterprise Agent v1.0.0 - CDP Edge Quantum Tier*