Merge pull request #366 from objectstack-ai/copilot/fix-action-step-error-again

Author: hotlong

examples/basic/ai-rag-example.ts

@@ -9,21 +9,14 @@
  * - AI-powered question answering
  */
 
-import type {
-  RAGPipelineConfig,
-  DocumentChunk,
-  DocumentMetadata,
-  RAGQueryRequest,
-  RAGQueryResponse,
-  Agent,
-} from '@objectstack/spec';
+import type { AI } from '@objectstack/spec';
 
 /**
  * Example 1: RAG Pipeline Configuration
  * 
  * Complete RAG pipeline setup for a knowledge base system
  */
-export const knowledgeBaseRAG: RAGPipelineConfig = {
+export const knowledgeBaseRAG: AI.RAGPipelineConfig = {
   name: 'knowledge_base_rag',
   label: 'Knowledge Base RAG Pipeline',
   description: 'RAG pipeline for company knowledge base',
@@ -43,6 +36,9 @@ export const knowledgeBaseRAG: RAGPipelineConfig = {
     namespace: 'production',
     dimensions: 1536,
     metric: 'cosine',
+    batchSize: 100,
+    connectionPoolSize: 10,
+    timeout: 30000,
   },
 
   // Chunking strategy
@@ -76,14 +72,18 @@ export const knowledgeBaseRAG: RAGPipelineConfig = {
   metadataFilters: {
     status: 'published',
   },
+
+  // Cache Configuration
+  enableCache: true,
+  cacheTTL: 3600,
 };
 
 /**
  * Example 2: Document Ingestion
  * 
  * How to index documents into the RAG pipeline
  */
-export const sampleDocumentMetadata: DocumentMetadata[] = [
+export const sampleDocumentMetadata: AI.DocumentMetadata[] = [
   {
     source: 'https://docs.objectstack.dev/architecture',
     sourceType: 'url',
@@ -110,7 +110,7 @@ export const sampleDocumentMetadata: DocumentMetadata[] = [
   },
 ];
 
-export const sampleDocumentChunks: DocumentChunk[] = [
+export const sampleDocumentChunks: AI.DocumentChunk[] = [
   {
     id: 'chunk_001',
     content: `ObjectStack is a metadata-driven low-code platform that enables rapid application development.
@@ -145,12 +145,14 @@ Use the Agent protocol to define AI assistants with specific capabilities.`,
  * 
  * Performing a RAG query with filters and options
  */
-export const sampleQueries: RAGQueryRequest[] = [
+export const sampleQueries: AI.RAGQueryRequest[] = [
   {
     // Simple question
     query: 'What is ObjectStack?',
     pipelineName: 'knowledge_base_rag',
     topK: 5,
+    includeMetadata: true,
+    includeSources: true,
   },
   {
     // Question with metadata filtering
@@ -160,6 +162,8 @@ export const sampleQueries: RAGQueryRequest[] = [
     metadataFilters: {
       category: 'Tutorial',
     },
+    includeMetadata: true,
+    includeSources: true,
   },
   {
     // Advanced query
@@ -179,7 +183,7 @@ export const sampleQueries: RAGQueryRequest[] = [
  * 
  * What the pipeline returns
  */
-export const sampleResults: RAGQueryResponse = {
+export const sampleResults: AI.RAGQueryResponse = {
   query: 'What is ObjectStack?',
 
   // Retrieved chunks
@@ -214,33 +218,21 @@ It uses a three-layer architecture: ObjectQL for data, ObjectUI for presentation
 
 [Source: https://docs.objectstack.dev/architecture]
 The platform supports multiple databases and provides built-in AI capabilities.`,
-
-  // Metadata
-  processingTimeMs: 150,
-  totalResults: 2,
 };
 
 /**
  * Example 5: AI Agent with RAG
  * 
  * Integrating RAG into an AI agent
  */
-export const ragEnabledAgent: Agent = {
+export const ragEnabledAgent: AI.Agent = {
   name: 'documentation_assistant',
-  type: 'conversational',
   label: 'Documentation Assistant',
-  description: 'AI assistant powered by RAG for answering questions about ObjectStack',
-
-  // Agent capabilities
-  capabilities: {
-    objectAccess: [],
-    canCreate: false,
-    canUpdate: false,
-    canAnalyze: true,
-  },
-
+  role: 'Documentation Support Specialist',
+  active: true,
+  
   // System prompt with RAG instructions
-  systemPrompt: `You are a helpful documentation assistant for ObjectStack.
+  instructions: `You are a helpful documentation assistant for ObjectStack.
 
 You have access to the knowledge base through RAG (Retrieval-Augmented Generation).
 When answering questions:
@@ -252,18 +244,9 @@ When answering questions:
 Always format your responses in a clear, structured way.`,
 
   // RAG Configuration
-  rag: {
-    pipeline: 'knowledge_base_rag',
-    enabled: true,
-    
-    // When to trigger RAG
-    trigger: 'always', // or 'auto', 'manual'
-    
-    // Number of chunks to retrieve
-    topK: 5,
-    
-    // Include source attribution
-    includeSources: true,
+  knowledge: {
+    indexes: ['knowledge-base'],
+    topics: ['ObjectStack', 'documentation'],
   },
 
   // Model configuration
@@ -277,12 +260,9 @@ Always format your responses in a clear, structured way.`,
   // Tools (optional - for function calling)
   tools: [
     {
+      type: 'query',
       name: 'search_documentation',
       description: 'Search the ObjectStack documentation for specific topics',
-      parameters: {
-        query: 'string',
-        category: 'string?',
-      },
     },
   ],
 };
@@ -304,7 +284,7 @@ export function demonstrateRAGUsage() {
   console.log('- Embedding query...');
   console.log('- Searching vector database...');
   console.log(`- Retrieved ${sampleResults.results.length} relevant chunks`);
-  console.log(`- Processing time: ${sampleResults.processingTimeMs}ms\n`);
+  console.log('- Processing complete\n');
 
   // Step 3: Context is assembled
   console.log('Assembled Context:');

examples/basic/api-discovery-example.ts

@@ -10,17 +10,14 @@
  * Typically exposed at: GET /api/discovery
  */
 
-import type {
-  DiscoveryResponse,
-  ApiCapabilities,
-} from '@objectstack/spec';
+import type { API } from '@objectstack/spec';
 
 /**
  * Example 1: Complete Discovery Response
  * 
  * This is what a client receives when calling /api/discovery
  */
-export const fullDiscoveryResponse: DiscoveryResponse = {
+export const fullDiscoveryResponse: API.DiscoveryResponse = {
   // System Identity
   name: 'ObjectStack CRM',
   version: '2.1.0',
@@ -57,7 +54,7 @@ export const fullDiscoveryResponse: DiscoveryResponse = {
  * 
  * A simplified response for local development
  */
-export const devDiscoveryResponse: DiscoveryResponse = {
+export const devDiscoveryResponse: API.DiscoveryResponse = {
   name: 'ObjectStack Dev',
   version: '0.1.0',
   environment: 'development',
@@ -88,12 +85,12 @@ export const devDiscoveryResponse: DiscoveryResponse = {
  * How a client can use the discovery API to adapt its behavior
  */
 export class AdaptiveClient {
-  private discovery: DiscoveryResponse | null = null;
+  private discovery: API.DiscoveryResponse | null = null;
 
   async initialize(baseUrl: string) {
     // Fetch discovery information
     const response = await fetch(`${baseUrl}/api/discovery`);
-    this.discovery = await response.json() as DiscoveryResponse;
+    this.discovery = await response.json() as API.DiscoveryResponse;
 
     console.log(`Connected to: ${this.discovery.name} v${this.discovery.version}`);
     console.log(`Environment: ${this.discovery.environment}`);
@@ -102,7 +99,7 @@ export class AdaptiveClient {
   /**
    * Check if a specific feature is available
    */
-  hasFeature(feature: keyof ApiCapabilities): boolean {
+  hasFeature(feature: keyof API.ApiCapabilities): boolean {
     if (!this.discovery) return false;
     return this.discovery.features[feature] === true;
   }
@@ -142,7 +139,7 @@ export class AdaptiveClient {
  * 
  * How an AI agent can use discovery to understand the system
  */
-export function generateSystemPromptFromDiscovery(discovery: DiscoveryResponse): string {
+export function generateSystemPromptFromDiscovery(discovery: API.DiscoveryResponse): string {
   const { name, version, features, locale } = discovery;
 
   const prompt = `You are an AI assistant for ${name} (v${version}).