@cmdoss/memwal-sdk 0.6.2 → 0.8.0

package/src/client/namespaces/consolidated/AINamespace.ts

@@ -1,449 +1,449 @@
-/**
- * AI Namespace - Consolidated AI Features
- *
- * Merges functionality from:
- * - EmbeddingsNamespace: Vector embedding generation
- * - ClassifyNamespace: Content classification & analysis
- *
- * Provides a unified interface for all AI-powered operations.
- *
- * @module client/namespaces/consolidated
- */
-
-import type { ServiceContainer } from '../../SimplePDWClient';
-
-// ============================================================================
-// Types
-// ============================================================================
-
-/**
- * Pattern analysis result
- */
-export interface PatternAnalysis {
-  patterns: Array<{
-    type: string;
-    confidence: number;
-    examples: string[];
-  }>;
-  categories: string[];
-  suggestedCategory: string;
-}
-
-/**
- * Classification result with full details
- */
-export interface ClassificationResult {
-  shouldSave: boolean;
-  confidence: number;
-  category: string;
-  reasoning: string;
-}
-
-/**
- * Embedding options
- */
-export interface EmbedOptions {
-  type?: 'query' | 'document';
-}
-
-// ============================================================================
-// AI Namespace
-// ============================================================================
-
-/**
- * AI Namespace - Unified AI Operations
- *
- * Consolidates embeddings and classification into one namespace.
- *
- * @example
- * ```typescript
- * // Generate embeddings
- * const vector = await pdw.ai.embed('Hello world');
- *
- * // Classify content
- * const category = await pdw.ai.classify('I love TypeScript');
- * ```
- */
-export class AINamespace {
-  constructor(private services: ServiceContainer) {}
-
-  // ==========================================================================
-  // Embedding Operations (from EmbeddingsNamespace)
-  // ==========================================================================
-
-  /**
-   * Generate embedding for single text
-   *
-   * @param text - Text to embed
-   * @param options - Embedding options
-   * @returns Embedding vector (768 dimensions for Gemini)
-   *
-   * @example
-   * ```typescript
-   * const vector = await pdw.ai.embed('My favorite color is blue');
-   * console.log(vector.length); // 768
-   * ```
-   */
-  async embed(text: string, options?: EmbedOptions): Promise<number[]> {
-    if (!this.services.embedding) {
-      throw new Error('Embedding service not configured. Please provide geminiApiKey.');
-    }
-    const result = await this.services.embedding.embedText({ text });
-    return result.vector;
-  }
-
-  /**
-   * Generate embeddings for multiple texts (batch)
-   *
-   * @param texts - Array of texts
-   * @returns Array of embedding vectors
-   *
-   * @example
-   * ```typescript
-   * const vectors = await pdw.ai.embedBatch(['Hello', 'World', 'TypeScript']);
-   * console.log(vectors.length); // 3
-   * ```
-   */
-  async embedBatch(texts: string[]): Promise<number[][]> {
-    if (!this.services.embedding) {
-      throw new Error('Embedding service not configured.');
-    }
-    const result = await this.services.embedding.embedBatch(texts);
-    return result.vectors;
-  }
-
-  /**
-   * Calculate cosine similarity between two vectors
-   *
-   * @param vector1 - First vector
-   * @param vector2 - Second vector
-   * @returns Similarity score (0-1, higher is more similar)
-   *
-   * @example
-   * ```typescript
-   * const v1 = await pdw.ai.embed('cat');
-   * const v2 = await pdw.ai.embed('dog');
-   * const similarity = pdw.ai.similarity(v1, v2);
-   * console.log(similarity); // ~0.85 (similar concepts)
-   * ```
-   */
-  similarity(vector1: number[], vector2: number[]): number {
-    if (!this.services.embedding) {
-      throw new Error('Embedding service not configured.');
-    }
-    return this.services.embedding.calculateCosineSimilarity(vector1, vector2);
-  }
-
-  /**
-   * Find most similar vectors from candidates
-   *
-   * @param queryVector - Query vector
-   * @param candidateVectors - Candidate vectors to compare
-   * @param k - Number of results (default: 5)
-   * @returns Top k similar vectors with scores
-   *
-   * @example
-   * ```typescript
-   * const query = await pdw.ai.embed('programming');
-   * const candidates = await pdw.ai.embedBatch(['TypeScript', 'cooking', 'Java']);
-   * const similar = pdw.ai.findSimilar(query, candidates, 2);
-   * // Returns indices 0 (TypeScript) and 2 (Java) as most similar
-   * ```
-   */
-  findSimilar(
-    queryVector: number[],
-    candidateVectors: number[][],
-    k: number = 5
-  ): Array<{ index: number; score: number }> {
-    if (!this.services.embedding) {
-      throw new Error('Embedding service not configured.');
-    }
-    const results = this.services.embedding.findMostSimilar(queryVector, candidateVectors, k);
-    return results.map(r => ({
-      index: r.index,
-      score: r.similarity
-    }));
-  }
-
-  // ==========================================================================
-  // Classification Operations (from ClassifyNamespace)
-  // ==========================================================================
-
-  /**
-   * Classify content into category
-   *
-   * Uses AI to determine the most appropriate category.
-   *
-   * @param content - Text content to classify
-   * @returns Category name (fact, preference, todo, note, general, etc.)
-   *
-   * @example
-   * ```typescript
-   * const category = await pdw.ai.classify('I prefer dark mode');
-   * // Returns: 'preference'
-   * ```
-   */
-  async classify(content: string): Promise<string> {
-    if (!this.services.classifier) {
-      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
-    }
-    return await this.services.classifier.classifyContent(content);
-  }
-
-  /**
-   * Determine if content should be saved as a memory
-   *
-   * Uses AI and pattern matching to decide if content is worth saving.
-   * By default, only explicit commands like "remember that" trigger saves.
-   *
-   * @param content - Text content to analyze
-   * @param options - Options for classification
-   * @param options.explicitOnly - If true (default), only explicit memory commands trigger save
-   * @returns true if should save, false otherwise
-   *
-   * @example
-   * ```typescript
-   * // Default: only explicit commands
-   * const shouldSave = await pdw.ai.shouldSave('Remember that I love TypeScript');
-   * // Returns: true
-   *
-   * // With explicitOnly=false, pattern matching is more aggressive
-   * const shouldSave2 = await pdw.ai.shouldSave('I love TypeScript', { explicitOnly: false });
-   * // Returns: true (matches "I love..." pattern)
-   * ```
-   */
-  async shouldSave(content: string, options?: { explicitOnly?: boolean }): Promise<boolean> {
-    if (!this.services.classifier) {
-      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
-    }
-    const result = await this.services.classifier.shouldSaveMemory(content, {
-      explicitOnly: options?.explicitOnly ?? true // Default to explicit-only
-    });
-    return result.shouldSave;
-  }
-
-  /**
-   * Extract memory content from explicit command
-   *
-   * Parses commands like "remember that X" and extracts X.
-   *
-   * @param message - User message to parse
-   * @returns Extracted content or null if no command found
-   *
-   * @example
-   * ```typescript
-   * const content = pdw.ai.extractMemoryContent('Remember that my name is John');
-   * // Returns: 'my name is John'
-   *
-   * const content2 = pdw.ai.extractMemoryContent('Hello there');
-   * // Returns: null
-   * ```
-   */
-  extractMemoryContent(message: string): string | null {
-    const patterns = [
-      /remember that\s+(.+)/i,
-      /remember:?\s+(.+)/i,
-      /don't forget that\s+(.+)/i,
-      /don't forget:?\s+(.+)/i,
-      /please remember\s+(.+)/i,
-      /store (?:this )?(?:in|to) memory:?\s*(.+)/i,
-      /save (?:this )?(?:in|to) memory:?\s*(.+)/i,
-      /add (?:this )?to (?:my )?memory:?\s*(.+)/i,
-      /keep in mind:?\s+(.+)/i,
-      /note that\s+(.+)/i,
-    ];
-
-    for (const pattern of patterns) {
-      const match = message.match(pattern);
-      if (match && match[1]) {
-        return match[1].trim();
-      }
-    }
-    return null;
-  }
-
-  /**
-   * Extract multiple memories from a single message
-   *
-   * Splits a message containing multiple facts/memories into individual items.
-   * Supports comma-separated, semicolon-separated, "and"-separated, and numbered lists.
-   *
-   * @param message - User message potentially containing multiple memories
-   * @returns Array of extracted memory contents, or empty array if no pattern matched
-   *
-   * @example
-   * ```typescript
-   * // Comma-separated
-   * pdw.ai.extractMultipleMemories('Remember that my name is John, I work at Google, and my birthday is Dec 25');
-   * // Returns: ['my name is John', 'I work at Google', 'my birthday is Dec 25']
-   *
-   * // Semicolon-separated
-   * pdw.ai.extractMultipleMemories('Note that: API key is abc123; server port is 3000; database is PostgreSQL');
-   * // Returns: ['API key is abc123', 'server port is 3000', 'database is PostgreSQL']
-   *
-   * // Numbered list
-   * pdw.ai.extractMultipleMemories('Remember: 1. My email is test@example 2. My phone is 123-456 3. I prefer dark mode');
-   * // Returns: ['My email is test@example', 'My phone is 123-456', 'I prefer dark mode']
-   *
-   * // Single memory (fallback to extractMemoryContent)
-   * pdw.ai.extractMultipleMemories('Remember that I like pizza');
-   * // Returns: ['I like pizza']
-   * ```
-   */
-  extractMultipleMemories(message: string): string[] {
-    // First, extract the content after the memory keyword
-    const content = this.extractMemoryContent(message);
-    if (!content) {
-      return [];
-    }
-
-    // Try numbered list pattern: "1. xxx 2. yyy 3. zzz"
-    const numberedPattern = /\d+\.\s*([^0-9]+?)(?=\d+\.|$)/g;
-    const numberedMatches = [...content.matchAll(numberedPattern)];
-    if (numberedMatches.length > 1) {
-      return numberedMatches
-        .map(m => m[1].trim())
-        .filter(s => s.length > 0);
-    }
-
-    // Try bullet list pattern: "- xxx - yyy" or "• xxx • yyy"
-    const bulletPattern = /[-•]\s*([^-•]+)/g;
-    const bulletMatches = [...content.matchAll(bulletPattern)];
-    if (bulletMatches.length > 1) {
-      return bulletMatches
-        .map(m => m[1].trim())
-        .filter(s => s.length > 0);
-    }
-
-    // Try semicolon-separated
-    if (content.includes(';')) {
-      const parts = content.split(';')
-        .map(s => s.trim())
-        .filter(s => s.length > 0);
-      if (parts.length > 1) {
-        return parts;
-      }
-    }
-
-    // Try comma + "and" pattern: "xxx, yyy, and zzz" or "xxx, yyy and zzz"
-    // Also handles simple comma-separated: "xxx, yyy, zzz"
-    const commaAndPattern = /,\s*(?:and\s+)?|\s+and\s+/i;
-    if (commaAndPattern.test(content)) {
-      const parts = content.split(commaAndPattern)
-        .map(s => s.trim())
-        .filter(s => s.length > 0);
-      if (parts.length > 1) {
-        return parts;
-      }
-    }
-
-    // Single memory - return as array with one item
-    return [content];
-  }
-
-  /**
-   * Check if message is a memory query (asking about stored memories)
-   *
-   * @param message - User message to check
-   * @returns true if user is asking about their memories
-   *
-   * @example
-   * ```typescript
-   * pdw.ai.isMemoryQuery('What do you remember about me?'); // true
-   * pdw.ai.isMemoryQuery('What is my name?'); // true
-   * pdw.ai.isMemoryQuery('Hello'); // false
-   * ```
-   */
-  isMemoryQuery(message: string): boolean {
-    const queryPatterns = [
-      /what do you (remember|know) about me/i,
-      /what('s| is) my (name|email|birthday|job|work|address)/i,
-      /do you (remember|know) my/i,
-      /tell me what you (remember|know)/i,
-      /what have i told you/i,
-      /what memories do you have/i,
-    ];
-    return queryPatterns.some(pattern => pattern.test(message));
-  }
-
-  /**
-   * Get full classification result with details
-   *
-   * @param content - Text content to classify
-   * @returns Full classification result with confidence and reasoning
-   *
-   * @example
-   * ```typescript
-   * const result = await pdw.ai.classifyFull('My birthday is January 15');
-   * // Returns: { shouldSave: true, confidence: 0.95, category: 'personal_info', reasoning: '...' }
-   * ```
-   */
-  async classifyFull(content: string): Promise<ClassificationResult> {
-    if (!this.services.classifier) {
-      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
-    }
-    return await this.services.classifier.shouldSaveMemory(content);
-  }
-
-  /**
-   * Analyze patterns in content
-   *
-   * Detects patterns like personal info, contact details, preferences, etc.
-   *
-   * @param content - Text content to analyze
-   * @returns Pattern analysis with detected patterns and suggested category
-   *
-   * @example
-   * ```typescript
-   * const analysis = await pdw.ai.patterns('My email is user@example.com');
-   * // Returns: { patterns: [{ type: 'contact', ... }], suggestedCategory: 'contact' }
-   * ```
-   */
-  async patterns(content: string): Promise<PatternAnalysis> {
-    if (!this.services.classifier) {
-      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
-    }
-    const result = await this.services.classifier.analyzePatterns(content);
-    return {
-      patterns: result.patterns.map((p: string) => ({
-        type: p,
-        confidence: 1.0,
-        examples: []
-      })),
-      categories: result.categories || [],
-      suggestedCategory: result.categories[0] || 'general'
-    };
-  }
-
-  /**
-   * Calculate importance score for content
-   *
-   * Uses AI to determine how important the content is (1-10 scale).
-   *
-   * @param content - Text content to score
-   * @returns Importance score (1-10)
-   *
-   * @example
-   * ```typescript
-   * const importance = await pdw.ai.importance('Emergency contact: 911');
-   * // Returns: 10
-   * ```
-   */
-  async importance(content: string): Promise<number> {
-    if (!this.services.classifier) {
-      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
-    }
-    const category = await this.services.classifier.classifyContent(content);
-    const categoryImportance: Record<string, number> = {
-      'emergency': 10,
-      'contact': 9,
-      'personal_info': 8,
-      'career': 7,
-      'preference': 6,
-      'fact': 5,
-      'note': 4,
-      'general': 3
-    };
-    return categoryImportance[category] || 5;
-  }
-}
+/**
+ * AI Namespace - Consolidated AI Features
+ *
+ * Merges functionality from:
+ * - EmbeddingsNamespace: Vector embedding generation
+ * - ClassifyNamespace: Content classification & analysis
+ *
+ * Provides a unified interface for all AI-powered operations.
+ *
+ * @module client/namespaces/consolidated
+ */
+
+import type { ServiceContainer } from '../../SimplePDWClient';
+
+// ============================================================================
+// Types
+// ============================================================================
+
+/**
+ * Pattern analysis result
+ */
+export interface PatternAnalysis {
+  patterns: Array<{
+    type: string;
+    confidence: number;
+    examples: string[];
+  }>;
+  categories: string[];
+  suggestedCategory: string;
+}
+
+/**
+ * Classification result with full details
+ */
+export interface ClassificationResult {
+  shouldSave: boolean;
+  confidence: number;
+  category: string;
+  reasoning: string;
+}
+
+/**
+ * Embedding options
+ */
+export interface EmbedOptions {
+  type?: 'query' | 'document';
+}
+
+// ============================================================================
+// AI Namespace
+// ============================================================================
+
+/**
+ * AI Namespace - Unified AI Operations
+ *
+ * Consolidates embeddings and classification into one namespace.
+ *
+ * @example
+ * ```typescript
+ * // Generate embeddings
+ * const vector = await pdw.ai.embed('Hello world');
+ *
+ * // Classify content
+ * const category = await pdw.ai.classify('I love TypeScript');
+ * ```
+ */
+export class AINamespace {
+  constructor(private services: ServiceContainer) {}
+
+  // ==========================================================================
+  // Embedding Operations (from EmbeddingsNamespace)
+  // ==========================================================================
+
+  /**
+   * Generate embedding for single text
+   *
+   * @param text - Text to embed
+   * @param options - Embedding options
+   * @returns Embedding vector (3072 dimensions for Gemini)
+   *
+   * @example
+   * ```typescript
+   * const vector = await pdw.ai.embed('My favorite color is blue');
+   * console.log(vector.length); // 3072
+   * ```
+   */
+  async embed(text: string, options?: EmbedOptions): Promise<number[]> {
+    if (!this.services.embedding) {
+      throw new Error('Embedding service not configured. Please provide geminiApiKey.');
+    }
+    const result = await this.services.embedding.embedText({ text });
+    return result.vector;
+  }
+
+  /**
+   * Generate embeddings for multiple texts (batch)
+   *
+   * @param texts - Array of texts
+   * @returns Array of embedding vectors
+   *
+   * @example
+   * ```typescript
+   * const vectors = await pdw.ai.embedBatch(['Hello', 'World', 'TypeScript']);
+   * console.log(vectors.length); // 3
+   * ```
+   */
+  async embedBatch(texts: string[]): Promise<number[][]> {
+    if (!this.services.embedding) {
+      throw new Error('Embedding service not configured.');
+    }
+    const result = await this.services.embedding.embedBatch(texts);
+    return result.vectors;
+  }
+
+  /**
+   * Calculate cosine similarity between two vectors
+   *
+   * @param vector1 - First vector
+   * @param vector2 - Second vector
+   * @returns Similarity score (0-1, higher is more similar)
+   *
+   * @example
+   * ```typescript
+   * const v1 = await pdw.ai.embed('cat');
+   * const v2 = await pdw.ai.embed('dog');
+   * const similarity = pdw.ai.similarity(v1, v2);
+   * console.log(similarity); // ~0.85 (similar concepts)
+   * ```
+   */
+  similarity(vector1: number[], vector2: number[]): number {
+    if (!this.services.embedding) {
+      throw new Error('Embedding service not configured.');
+    }
+    return this.services.embedding.calculateCosineSimilarity(vector1, vector2);
+  }
+
+  /**
+   * Find most similar vectors from candidates
+   *
+   * @param queryVector - Query vector
+   * @param candidateVectors - Candidate vectors to compare
+   * @param k - Number of results (default: 5)
+   * @returns Top k similar vectors with scores
+   *
+   * @example
+   * ```typescript
+   * const query = await pdw.ai.embed('programming');
+   * const candidates = await pdw.ai.embedBatch(['TypeScript', 'cooking', 'Java']);
+   * const similar = pdw.ai.findSimilar(query, candidates, 2);
+   * // Returns indices 0 (TypeScript) and 2 (Java) as most similar
+   * ```
+   */
+  findSimilar(
+    queryVector: number[],
+    candidateVectors: number[][],
+    k: number = 5
+  ): Array<{ index: number; score: number }> {
+    if (!this.services.embedding) {
+      throw new Error('Embedding service not configured.');
+    }
+    const results = this.services.embedding.findMostSimilar(queryVector, candidateVectors, k);
+    return results.map(r => ({
+      index: r.index,
+      score: r.similarity
+    }));
+  }
+
+  // ==========================================================================
+  // Classification Operations (from ClassifyNamespace)
+  // ==========================================================================
+
+  /**
+   * Classify content into category
+   *
+   * Uses AI to determine the most appropriate category.
+   *
+   * @param content - Text content to classify
+   * @returns Category name (fact, preference, todo, note, general, etc.)
+   *
+   * @example
+   * ```typescript
+   * const category = await pdw.ai.classify('I prefer dark mode');
+   * // Returns: 'preference'
+   * ```
+   */
+  async classify(content: string): Promise<string> {
+    if (!this.services.classifier) {
+      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
+    }
+    return await this.services.classifier.classifyContent(content);
+  }
+
+  /**
+   * Determine if content should be saved as a memory
+   *
+   * Uses AI and pattern matching to decide if content is worth saving.
+   * By default, only explicit commands like "remember that" trigger saves.
+   *
+   * @param content - Text content to analyze
+   * @param options - Options for classification
+   * @param options.explicitOnly - If true (default), only explicit memory commands trigger save
+   * @returns true if should save, false otherwise
+   *
+   * @example
+   * ```typescript
+   * // Default: only explicit commands
+   * const shouldSave = await pdw.ai.shouldSave('Remember that I love TypeScript');
+   * // Returns: true
+   *
+   * // With explicitOnly=false, pattern matching is more aggressive
+   * const shouldSave2 = await pdw.ai.shouldSave('I love TypeScript', { explicitOnly: false });
+   * // Returns: true (matches "I love..." pattern)
+   * ```
+   */
+  async shouldSave(content: string, options?: { explicitOnly?: boolean }): Promise<boolean> {
+    if (!this.services.classifier) {
+      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
+    }
+    const result = await this.services.classifier.shouldSaveMemory(content, {
+      explicitOnly: options?.explicitOnly ?? true // Default to explicit-only
+    });
+    return result.shouldSave;
+  }
+
+  /**
+   * Extract memory content from explicit command
+   *
+   * Parses commands like "remember that X" and extracts X.
+   *
+   * @param message - User message to parse
+   * @returns Extracted content or null if no command found
+   *
+   * @example
+   * ```typescript
+   * const content = pdw.ai.extractMemoryContent('Remember that my name is John');
+   * // Returns: 'my name is John'
+   *
+   * const content2 = pdw.ai.extractMemoryContent('Hello there');
+   * // Returns: null
+   * ```
+   */
+  extractMemoryContent(message: string): string | null {
+    const patterns = [
+      /remember that\s+(.+)/i,
+      /remember:?\s+(.+)/i,
+      /don't forget that\s+(.+)/i,
+      /don't forget:?\s+(.+)/i,
+      /please remember\s+(.+)/i,
+      /store (?:this )?(?:in|to) memory:?\s*(.+)/i,
+      /save (?:this )?(?:in|to) memory:?\s*(.+)/i,
+      /add (?:this )?to (?:my )?memory:?\s*(.+)/i,
+      /keep in mind:?\s+(.+)/i,
+      /note that\s+(.+)/i,
+    ];
+
+    for (const pattern of patterns) {
+      const match = message.match(pattern);
+      if (match && match[1]) {
+        return match[1].trim();
+      }
+    }
+    return null;
+  }
+
+  /**
+   * Extract multiple memories from a single message
+   *
+   * Splits a message containing multiple facts/memories into individual items.
+   * Supports comma-separated, semicolon-separated, "and"-separated, and numbered lists.
+   *
+   * @param message - User message potentially containing multiple memories
+   * @returns Array of extracted memory contents, or empty array if no pattern matched
+   *
+   * @example
+   * ```typescript
+   * // Comma-separated
+   * pdw.ai.extractMultipleMemories('Remember that my name is John, I work at Google, and my birthday is Dec 25');
+   * // Returns: ['my name is John', 'I work at Google', 'my birthday is Dec 25']
+   *
+   * // Semicolon-separated
+   * pdw.ai.extractMultipleMemories('Note that: API key is abc123; server port is 3000; database is PostgreSQL');
+   * // Returns: ['API key is abc123', 'server port is 3000', 'database is PostgreSQL']
+   *
+   * // Numbered list
+   * pdw.ai.extractMultipleMemories('Remember: 1. My email is test@example 2. My phone is 123-456 3. I prefer dark mode');
+   * // Returns: ['My email is test@example', 'My phone is 123-456', 'I prefer dark mode']
+   *
+   * // Single memory (fallback to extractMemoryContent)
+   * pdw.ai.extractMultipleMemories('Remember that I like pizza');
+   * // Returns: ['I like pizza']
+   * ```
+   */
+  extractMultipleMemories(message: string): string[] {
+    // First, extract the content after the memory keyword
+    const content = this.extractMemoryContent(message);
+    if (!content) {
+      return [];
+    }
+
+    // Try numbered list pattern: "1. xxx 2. yyy 3. zzz"
+    const numberedPattern = /\d+\.\s*([^0-9]+?)(?=\d+\.|$)/g;
+    const numberedMatches = [...content.matchAll(numberedPattern)];
+    if (numberedMatches.length > 1) {
+      return numberedMatches
+        .map(m => m[1].trim())
+        .filter(s => s.length > 0);
+    }
+
+    // Try bullet list pattern: "- xxx - yyy" or "• xxx • yyy"
+    const bulletPattern = /[-•]\s*([^-•]+)/g;
+    const bulletMatches = [...content.matchAll(bulletPattern)];
+    if (bulletMatches.length > 1) {
+      return bulletMatches
+        .map(m => m[1].trim())
+        .filter(s => s.length > 0);
+    }
+
+    // Try semicolon-separated
+    if (content.includes(';')) {
+      const parts = content.split(';')
+        .map(s => s.trim())
+        .filter(s => s.length > 0);
+      if (parts.length > 1) {
+        return parts;
+      }
+    }
+
+    // Try comma + "and" pattern: "xxx, yyy, and zzz" or "xxx, yyy and zzz"
+    // Also handles simple comma-separated: "xxx, yyy, zzz"
+    const commaAndPattern = /,\s*(?:and\s+)?|\s+and\s+/i;
+    if (commaAndPattern.test(content)) {
+      const parts = content.split(commaAndPattern)
+        .map(s => s.trim())
+        .filter(s => s.length > 0);
+      if (parts.length > 1) {
+        return parts;
+      }
+    }
+
+    // Single memory - return as array with one item
+    return [content];
+  }
+
+  /**
+   * Check if message is a memory query (asking about stored memories)
+   *
+   * @param message - User message to check
+   * @returns true if user is asking about their memories
+   *
+   * @example
+   * ```typescript
+   * pdw.ai.isMemoryQuery('What do you remember about me?'); // true
+   * pdw.ai.isMemoryQuery('What is my name?'); // true
+   * pdw.ai.isMemoryQuery('Hello'); // false
+   * ```
+   */
+  isMemoryQuery(message: string): boolean {
+    const queryPatterns = [
+      /what do you (remember|know) about me/i,
+      /what('s| is) my (name|email|birthday|job|work|address)/i,
+      /do you (remember|know) my/i,
+      /tell me what you (remember|know)/i,
+      /what have i told you/i,
+      /what memories do you have/i,
+    ];
+    return queryPatterns.some(pattern => pattern.test(message));
+  }
+
+  /**
+   * Get full classification result with details
+   *
+   * @param content - Text content to classify
+   * @returns Full classification result with confidence and reasoning
+   *
+   * @example
+   * ```typescript
+   * const result = await pdw.ai.classifyFull('My birthday is January 15');
+   * // Returns: { shouldSave: true, confidence: 0.95, category: 'personal_info', reasoning: '...' }
+   * ```
+   */
+  async classifyFull(content: string): Promise<ClassificationResult> {
+    if (!this.services.classifier) {
+      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
+    }
+    return await this.services.classifier.shouldSaveMemory(content);
+  }
+
+  /**
+   * Analyze patterns in content
+   *
+   * Detects patterns like personal info, contact details, preferences, etc.
+   *
+   * @param content - Text content to analyze
+   * @returns Pattern analysis with detected patterns and suggested category
+   *
+   * @example
+   * ```typescript
+   * const analysis = await pdw.ai.patterns('My email is user@example.com');
+   * // Returns: { patterns: [{ type: 'contact', ... }], suggestedCategory: 'contact' }
+   * ```
+   */
+  async patterns(content: string): Promise<PatternAnalysis> {
+    if (!this.services.classifier) {
+      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
+    }
+    const result = await this.services.classifier.analyzePatterns(content);
+    return {
+      patterns: result.patterns.map((p: string) => ({
+        type: p,
+        confidence: 1.0,
+        examples: []
+      })),
+      categories: result.categories || [],
+      suggestedCategory: result.categories[0] || 'general'
+    };
+  }
+
+  /**
+   * Calculate importance score for content
+   *
+   * Uses AI to determine how important the content is (1-10 scale).
+   *
+   * @param content - Text content to score
+   * @returns Importance score (1-10)
+   *
+   * @example
+   * ```typescript
+   * const importance = await pdw.ai.importance('Emergency contact: 911');
+   * // Returns: 10
+   * ```
+   */
+  async importance(content: string): Promise<number> {
+    if (!this.services.classifier) {
+      throw new Error('Classifier service not configured. Please provide geminiApiKey.');
+    }
+    const category = await this.services.classifier.classifyContent(content);
+    const categoryImportance: Record<string, number> = {
+      'emergency': 10,
+      'contact': 9,
+      'personal_info': 8,
+      'career': 7,
+      'preference': 6,
+      'fact': 5,
+      'note': 4,
+      'general': 3
+    };
+    return categoryImportance[category] || 5;
+  }
+}