♻️ refactor(chat): simplify chat service and update seed data

- Remove unused code from chat.service.ts - Simplify seed.ts with fewer AI models - Update CLAUDE.md documentation - Update .env.example - Remove unused package dependency - Minor UI fix in chat page
2026-05-14 19:41:09 +02:00 · 2025-12-12 20:48:10 +01:00 · 2025-12-12 20:48:10 +01:00 · 3f9bc5761b
commit 3f9bc5761b
parent 0fa154c7d6
6 changed files with 75 additions and 393 deletions
--- a/apps/chat/CLAUDE.md
+++ b/apps/chat/CLAUDE.md
@ -24,6 +24,7 @@ pnpm dev:chat:mobile             # Start mobile app
 pnpm dev:chat:web                # Start web app
 pnpm dev:chat:landing            # Start landing page
 pnpm dev:chat:backend            # Start backend server
+pnpm dev:chat:full               # Start backend + web + auth together
 ```

 ### Mobile App (chat/apps/mobile)
@ -43,6 +44,9 @@ pnpm build:prod                  # Build production version
 pnpm start:dev                   # Start with hot reload
 pnpm build                       # Build for production
 pnpm start:prod                  # Start production server
+pnpm db:push                     # Push schema to database
+pnpm db:seed                     # Seed AI models
+pnpm db:studio                   # Open Drizzle Studio
 ```

 ### Web App (chat/apps/web)
@ -66,7 +70,8 @@ pnpm preview                     # Preview production build
 - **Mobile**: React Native 0.76.7 + Expo SDK 52, NativeWind, Expo Router
 - **Web**: SvelteKit 2.x, Svelte 5, Tailwind CSS 4
 - **Landing**: Astro 5.16, Tailwind CSS
- **Backend**: NestJS 10, OpenRouter/Gemini AI, Supabase
+- **Backend**: NestJS 10, OpenRouter AI, Drizzle ORM, PostgreSQL
+- **Auth**: Mana Core Auth (JWT)
 - **Types**: TypeScript 5.x

 ## Architecture
@ -75,34 +80,45 @@ pnpm preview                     # Preview production build

 | Endpoint                          | Method | Description                 |
 | --------------------------------- | ------ | --------------------------- |
-| `/api/health`                     | GET    | Health check                |
-| `/api/chat/models`                | GET    | List available AI models    |
-| `/api/chat/completions`           | POST   | Create chat completion      |
-| `/api/conversations`              | GET    | List user conversations     |
-| `/api/conversations/:id`          | GET    | Get conversation details    |
-| `/api/conversations/:id/messages` | GET    | Get conversation messages   |
-| `/api/conversations`              | POST   | Create new conversation     |
-| `/api/conversations/:id/messages` | POST   | Add message to conversation |
+| `/api/v1/health`                  | GET    | Health check                |
+| `/api/v1/chat/models`             | GET    | List available AI models    |
+| `/api/v1/chat/completions`        | POST   | Create chat completion      |
+| `/api/v1/conversations`           | GET    | List user conversations     |
+| `/api/v1/conversations/:id`       | GET    | Get conversation details    |
+| `/api/v1/conversations/:id/messages` | GET | Get conversation messages   |
+| `/api/v1/conversations`           | POST   | Create new conversation     |
+| `/api/v1/conversations/:id/messages` | POST | Add message to conversation |

 ### Environment Variables

 #### Backend (.env)

-```
-OPENROUTER_API_KEY=...          # Get at https://openrouter.ai/keys
-GOOGLE_GENAI_API_KEY=...        # Optional: For Gemini models
-SUPABASE_URL=https://...
-SUPABASE_SERVICE_KEY=...
+```env
+# Required - All AI models via OpenRouter
+OPENROUTER_API_KEY=sk-or-v1-xxx    # Get at https://openrouter.ai/keys
+
+# Database (uses shared Docker PostgreSQL)
+DATABASE_URL=postgresql://manacore:devpassword@localhost:5432/chat
+
+# Auth
+MANA_CORE_AUTH_URL=http://localhost:3001
+
+# Server
 PORT=3002
-DEV_BYPASS_AUTH=true            # Optional: Skip auth in development
 ```

 #### Mobile (.env)

+```env
+EXPO_PUBLIC_MANA_CORE_AUTH_URL=http://localhost:3001
+EXPO_PUBLIC_BACKEND_URL=http://localhost:3002
 ```
-EXPO_PUBLIC_SUPABASE_URL=https://...
-EXPO_PUBLIC_SUPABASE_ANON_KEY=...
-EXPO_PUBLIC_BACKEND_URL=http://localhost:3001
+
+#### Web (.env)
+
+```env
+PUBLIC_MANA_CORE_AUTH_URL=http://localhost:3001
+PUBLIC_BACKEND_URL=http://localhost:3002
 ```

 ## Code Style Guidelines
@ -113,39 +129,43 @@ EXPO_PUBLIC_BACKEND_URL=http://localhost:3001
 - **Styling**: Tailwind CSS everywhere
 - **Formatting**: 100 char line limit, 2 space tabs, single quotes

-## AI Models Available
+## AI Models Available (via OpenRouter)

-### OpenRouter Models (Recommended)
+All models are accessed through OpenRouter, providing access to 100+ models with a single API key.

 | Model ID | Name | Price | Best For |
 | -------- | ---- | ----- | -------- |
-| ...440201 | Llama 3.1 8B | $0.05/M | Everyday tasks, cheap |
+| ...440201 | Llama 3.1 8B | $0.05/M | Everyday tasks (default) |
 | ...440202 | Llama 3.1 70B | $0.35/M | Complex reasoning |
 | ...440203 | DeepSeek V3 | $0.14/M | Reasoning at low cost |
 | ...440204 | Mistral Small | $0.10/M | General tasks |
 | ...440205 | Claude 3.5 Sonnet | $3/M | Best quality |
 | ...440206 | GPT-4o Mini | $0.15/M | Balanced performance |

-### Google Gemini Models
+## Quick Start

-| Model ID | Name | Description | Default |
-| -------- | ---- | ----------- | ------- |
-| ...440101 | Gemini 2.5 Flash | Fast, efficient responses | Yes |
-| ...440102 | Gemini 2.0 Flash-Lite | Ultra-lightweight model | No |
-| ...440103 | Gemini 2.5 Pro | Most capable model | No |
-
-## OpenRouter Setup
-
-To enable OpenRouter models:
-
- [ ] Get API key at https://openrouter.ai/keys
- [ ] Add `OPENROUTER_API_KEY=sk-or-v1-xxx` to `apps/chat/apps/backend/.env`
- [ ] Re-seed database: `pnpm --filter @chat/backend db:seed`
- [ ] Test: `pnpm dev:chat:backend`
+1. **Get OpenRouter API key** at https://openrouter.ai/keys
+2. **Create `.env`** in `apps/chat/apps/backend/`:
+   ```env
+   OPENROUTER_API_KEY=sk-or-v1-xxx
+   DATABASE_URL=postgresql://manacore:devpassword@localhost:5432/chat
+   MANA_CORE_AUTH_URL=http://localhost:3001
+   PORT=3002
+   ```
+3. **Start services**:
+   ```bash
+   pnpm docker:up              # Start PostgreSQL
+   pnpm dev:chat:full          # Start auth + backend + web
+   ```
+4. **Seed database** (first time only):
+   ```bash
+   pnpm --filter @chat/backend db:push
+   pnpm --filter @chat/backend db:seed
+   ```

 ## Important Notes

 1. **Security**: API keys are stored in the backend only - never in client apps
-2. **Authentication**: Uses Supabase Auth, shared with Mana Core ecosystem
-3. **Database**: Supabase PostgreSQL with RLS policies
-4. **Deployment**: Backend runs on port 3001 by default
+2. **Authentication**: Uses Mana Core Auth (JWT tokens)
+3. **Database**: PostgreSQL with Drizzle ORM (uses shared Docker container)
+4. **Deployment**: Backend runs on port 3002
--- a/apps/chat/apps/backend/.env.example
+++ b/apps/chat/apps/backend/.env.example
@ -1,20 +1,14 @@
-# OpenRouter Configuration (Recommended - multi-model access)
+# OpenRouter Configuration (Required)
 # Get your API key at https://openrouter.ai/keys
+# All AI models are accessed through OpenRouter
 OPENROUTER_API_KEY=your-openrouter-api-key

-# Google Gemini Configuration
-GOOGLE_GENAI_API_KEY=your-google-api-key
-
-# Azure OpenAI Configuration (Optional)
-AZURE_OPENAI_ENDPOINT=https://your-azure-openai-endpoint.openai.azure.com
-AZURE_OPENAI_API_KEY=your-api-key-here
-AZURE_OPENAI_API_VERSION=2024-12-01-preview
-
 # Mana Core Auth Configuration
 MANA_CORE_AUTH_URL=http://localhost:3001

 # PostgreSQL Database Configuration
-DATABASE_URL=postgresql://chat:password@localhost:5432/chat
+# Uses shared Docker PostgreSQL with separate 'chat' database
+DATABASE_URL=postgresql://manacore:devpassword@localhost:5432/chat

 # Server Configuration
 PORT=3002
--- a/apps/chat/apps/backend/package.json
+++ b/apps/chat/apps/backend/package.json
@ -24,7 +24,6 @@
 		"docker:clean": "docker compose down -v --rmi local"
 	},
 	"dependencies": {
-		"@google/generative-ai": "^0.24.1",
 		"@manacore/shared-errors": "workspace:*",
 		"@manacore/shared-nestjs-auth": "workspace:*",
 		"@nestjs/common": "^10.4.15",
--- a/apps/chat/apps/backend/src/chat/chat.service.ts
+++ b/apps/chat/apps/backend/src/chat/chat.service.ts
@ -2,7 +2,6 @@ import { Injectable, Inject, Logger } from '@nestjs/common';
 import { ConfigService } from '@nestjs/config';
 import { eq } from 'drizzle-orm';
 import { AsyncResult, ok, err, ValidationError, ServiceError } from '@manacore/shared-errors';
-import { GoogleGenerativeAI } from '@google/generative-ai';
 import OpenAI from 'openai';
 import { DATABASE_CONNECTION } from '../db/database.module';
 import { Database } from '../db/connection';
@ -14,37 +13,14 @@ import type { ChatCompletionResponseDto } from './dto/chat-completion.dto';
@Injectable()
 export class ChatService {
 	private readonly logger = new Logger(ChatService.name);
-	// Azure OpenAI config
-	private readonly azureApiKey: string;
-	private readonly azureEndpoint: string;
-	private readonly azureApiVersion: string;
-	// Google Gemini config
-	private readonly geminiClient: GoogleGenerativeAI | null = null;
-	// OpenRouter config
+	// OpenRouter config (primary provider)
 	private readonly openRouterClient: OpenAI | null = null;

 	constructor(
 		private configService: ConfigService,
 		@Inject(DATABASE_CONNECTION) private readonly db: Database
 	) {
-		// Azure OpenAI setup
-		this.azureApiKey = this.configService.get<string>('AZURE_OPENAI_API_KEY') || '';
-		this.azureEndpoint =
-			this.configService.get<string>('AZURE_OPENAI_ENDPOINT') ||
-			'https://memoroseopenai.openai.azure.com';
-		this.azureApiVersion =
-			this.configService.get<string>('AZURE_OPENAI_API_VERSION') || '2024-12-01-preview';
-
-		// Google Gemini setup
-		const geminiApiKey = this.configService.get<string>('GOOGLE_GENAI_API_KEY');
-		if (geminiApiKey) {
-			this.geminiClient = new GoogleGenerativeAI(geminiApiKey);
-			this.logger.log('Google Gemini client initialized');
-		} else {
-			this.logger.warn('GOOGLE_GENAI_API_KEY is not set - Gemini models unavailable');
-		}
-
-		// OpenRouter setup
+		// OpenRouter setup (primary and only provider)
 		const openRouterApiKey = this.configService.get<string>('OPENROUTER_API_KEY');
 		if (openRouterApiKey) {
 			this.openRouterClient = new OpenAI({
@ -57,11 +33,7 @@ export class ChatService {
 			});
 			this.logger.log('OpenRouter client initialized');
 		} else {
-			this.logger.warn('OPENROUTER_API_KEY is not set - OpenRouter models unavailable');
-		}
-
-		if (!this.azureApiKey) {
-			this.logger.warn('AZURE_OPENAI_API_KEY is not set - Azure models unavailable');
+			this.logger.error('OPENROUTER_API_KEY is not set - Chat will not work!');
 		}
 	}

@ -100,176 +72,8 @@ export class ChatService {
 			this.logger.log(`User ${userId} creating chat completion with model ${dto.modelId}`);
 		}

-		// Route to appropriate provider
-		if (model.provider === 'gemini') {
-			return this.createGeminiCompletion(model, dto);
-		} else if (model.provider === 'openrouter') {
-			return this.createOpenRouterCompletion(model, dto);
-		} else {
-			return this.createAzureCompletion(model, dto);
-		}
-	}
-
-	private async createGeminiCompletion(
-		model: Model,
-		dto: ChatCompletionDto
-	): AsyncResult<ChatCompletionResponseDto> {
-		if (!this.geminiClient) {
-			return err(ServiceError.externalError('Google Gemini', 'Gemini client not configured'));
-		}
-
-		const params = model.parameters as {
-			model?: string;
-			temperature?: number;
-			max_tokens?: number;
-		} | null;
-
-		const modelName = params?.model || 'gemini-2.5-flash';
-		const temperature = dto.temperature ?? params?.temperature ?? 0.7;
-		const maxTokens = dto.maxTokens ?? params?.max_tokens ?? 8192;
-
-		this.logger.log(`Sending request to Google Gemini model: ${modelName}`);
-
-		try {
-			const genModel = this.geminiClient.getGenerativeModel({
-				model: modelName,
-				generationConfig: {
-					temperature,
-					maxOutputTokens: maxTokens,
-				},
-			});
-
-			// Convert messages to Gemini format
-			// Gemini expects alternating user/model messages, with system as first user message
-			const systemMessages = dto.messages.filter((m) => m.role === 'system');
-			const chatMessages = dto.messages.filter((m) => m.role !== 'system');
-
-			// Build history for chat (all but last message)
-			const history = chatMessages.slice(0, -1).map((msg) => ({
-				role: msg.role === 'user' ? 'user' : 'model',
-				parts: [{ text: msg.content }],
-			}));
-
-			// Last message to send
-			const lastMessage = chatMessages[chatMessages.length - 1];
-			let userPrompt = lastMessage?.content || '';
-
-			// Prepend system instruction if present
-			if (systemMessages.length > 0) {
-				const systemPrompt = systemMessages.map((m) => m.content).join('\n');
-				userPrompt = `${systemPrompt}\n\n${userPrompt}`;
-			}
-
-			const chat = genModel.startChat({ history });
-			const result = await chat.sendMessage(userPrompt);
-			const response = result.response;
-			const messageContent = response.text();
-
-			if (!messageContent) {
-				this.logger.warn('No message content in Gemini response');
-				return err(ServiceError.generationFailed('Google Gemini', 'No response generated'));
-			}
-
-			// Gemini provides usage metadata
-			const usageMetadata = response.usageMetadata;
-
-			return ok({
-				content: messageContent,
-				usage: {
-					prompt_tokens: usageMetadata?.promptTokenCount || 0,
-					completion_tokens: usageMetadata?.candidatesTokenCount || 0,
-					total_tokens: usageMetadata?.totalTokenCount || 0,
-				},
-			});
-		} catch (error) {
-			this.logger.error('Error calling Google Gemini API', error);
-			return err(
-				ServiceError.generationFailed(
-					'Google Gemini',
-					error instanceof Error ? error.message : 'Unknown error',
-					error instanceof Error ? error : undefined
-				)
-			);
-		}
-	}
-
-	private async createAzureCompletion(
-		model: Model,
-		dto: ChatCompletionDto
-	): AsyncResult<ChatCompletionResponseDto> {
-		const params = model.parameters as {
-			deployment?: string;
-			temperature?: number;
-			max_tokens?: number;
-		} | null;
-
-		const deployment = params?.deployment || 'gpt-4o-mini-se';
-		const temperature = dto.temperature ?? params?.temperature ?? 0.7;
-		const maxTokens = dto.maxTokens ?? params?.max_tokens ?? 1000;
-
-		// Prepare request body
-		const requestBody: Record<string, unknown> = {
-			messages: dto.messages.map((msg) => ({
-				role: msg.role,
-				content: msg.content,
-			})),
-		};
-
-		// Model-specific parameters
-		const isGPTOModel = deployment.includes('gpt-o') || deployment.includes('gpt-4o');
-
-		if (!isGPTOModel) {
-			requestBody.max_tokens = maxTokens;
-			requestBody.temperature = temperature;
-		}
-
-		const url = `${this.azureEndpoint}/openai/deployments/${deployment}/chat/completions?api-version=${this.azureApiVersion}`;
-
-		this.logger.log(`Sending request to Azure OpenAI: ${url}`);
-
-		try {
-			const response = await fetch(url, {
-				method: 'POST',
-				headers: {
-					'Content-Type': 'application/json',
-					'api-key': this.azureApiKey,
-				},
-				body: JSON.stringify(requestBody),
-			});
-
-			if (!response.ok) {
-				const errorText = await response.text();
-				this.logger.error(`API error: ${response.status} - ${errorText}`);
-				return err(ServiceError.externalError('Azure OpenAI', `API error: ${response.status}`));
-			}
-
-			const data = await response.json();
-
-			const messageContent = data.choices?.[0]?.message?.content;
-
-			if (!messageContent) {
-				this.logger.warn('No message content in response');
-				return err(ServiceError.generationFailed('Azure OpenAI', 'No response generated'));
-			}
-
-			return ok({
-				content: messageContent,
-				usage: {
-					prompt_tokens: data.usage?.prompt_tokens || 0,
-					completion_tokens: data.usage?.completion_tokens || 0,
-					total_tokens: data.usage?.total_tokens || 0,
-				},
-			});
-		} catch (error) {
-			this.logger.error('Error calling Azure OpenAI API', error);
-			return err(
-				ServiceError.generationFailed(
-					'Azure OpenAI',
-					error instanceof Error ? error.message : 'Unknown error',
-					error instanceof Error ? error : undefined
-				)
-			);
-		}
+		// All models go through OpenRouter
+		return this.createOpenRouterCompletion(model, dto);
 	}

 	private async createOpenRouterCompletion(
--- a/apps/chat/apps/backend/src/db/seed.ts
+++ b/apps/chat/apps/backend/src/db/seed.ts
@ -33,49 +33,7 @@ async function seed() {

 		const modelData = [
 			// ============================================
-			// Google Gemini Models (Primary - fast & cost-effective)
-			// ============================================
-			{
-				id: '550e8400-e29b-41d4-a716-446655440101',
-				name: 'Gemini 2.5 Flash',
-				description: 'Fastest & most cost-effective - ideal for everyday tasks',
-				provider: 'gemini',
-				parameters: {
-					model: 'gemini-2.5-flash',
-					temperature: 0.7,
-					max_tokens: 8192,
-				},
-				isActive: true,
-				isDefault: true, // Default model
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440102',
-				name: 'Gemini 2.0 Flash-Lite',
-				description: 'Ultra-fast lightweight model - minimal latency',
-				provider: 'gemini',
-				parameters: {
-					model: 'gemini-2.0-flash-lite',
-					temperature: 0.7,
-					max_tokens: 4096,
-				},
-				isActive: true,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440103',
-				name: 'Gemini 2.5 Pro',
-				description: 'Most powerful Gemini - complex reasoning & analysis',
-				provider: 'gemini',
-				parameters: {
-					model: 'gemini-2.5-pro',
-					temperature: 0.7,
-					max_tokens: 16384,
-				},
-				isActive: true,
-				isDefault: false,
-			},
-			// ============================================
-			// OpenRouter Models (Multi-provider, cost-effective)
+			// OpenRouter Models (All models via OpenRouter)
 			// ============================================
 			{
 				id: '550e8400-e29b-41d4-a716-446655440201',
@ -88,7 +46,7 @@ async function seed() {
 					max_tokens: 4096,
 				},
 				isActive: true,
-				isDefault: false,
+				isDefault: true, // Default model - fast and cost-effective
 			},
 			{
 				id: '550e8400-e29b-41d4-a716-446655440202',
@ -155,101 +113,6 @@ async function seed() {
 				isActive: true,
 				isDefault: false,
 			},
-			// ============================================
-			// Azure OpenAI GPT-5 Family (Inactive - no deployment)
-			// ============================================
-			{
-				id: '550e8400-e29b-41d4-a716-446655440001',
-				name: 'GPT-5 Mini',
-				description: 'Fast & cost-effective - best for everyday tasks',
-				provider: 'azure',
-				parameters: {
-					temperature: 0.7,
-					max_tokens: 8192,
-					deployment: 'gpt-5-mini',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440002',
-				name: 'GPT-5 Nano',
-				description: 'Ultra-fast responses with low latency',
-				provider: 'azure',
-				parameters: {
-					temperature: 0.7,
-					max_tokens: 4096,
-					deployment: 'gpt-5-nano',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440003',
-				name: 'GPT-5 Chat',
-				description: 'Advanced multimodal conversations with emotional intelligence',
-				provider: 'azure',
-				parameters: {
-					temperature: 0.7,
-					max_tokens: 16384,
-					deployment: 'gpt-5-chat',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440004',
-				name: 'GPT-5',
-				description: 'Most powerful LLM - logic-heavy & multi-step tasks',
-				provider: 'azure',
-				parameters: {
-					temperature: 0.7,
-					max_tokens: 32768,
-					deployment: 'gpt-5',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440005',
-				name: 'GPT-5 Codex',
-				description: 'Optimized for coding & front-end development',
-				provider: 'azure',
-				parameters: {
-					temperature: 0.7,
-					max_tokens: 32768,
-					deployment: 'gpt-5-codex',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			// O-Series Reasoning Models (Inactive - no deployment)
-			{
-				id: '550e8400-e29b-41d4-a716-446655440006',
-				name: 'o4-mini',
-				description: 'Latest reasoning model - best for STEM & code',
-				provider: 'azure',
-				parameters: {
-					temperature: 1, // Reasoning models work best with temp=1
-					max_tokens: 16384,
-					deployment: 'o4-mini',
-				},
-				isActive: false,
-				isDefault: false,
-			},
-			{
-				id: '550e8400-e29b-41d4-a716-446655440007',
-				name: 'o3',
-				description: 'Advanced reasoning - 20% fewer errors than o1',
-				provider: 'azure',
-				parameters: {
-					temperature: 1,
-					max_tokens: 32768,
-					deployment: 'o3',
-				},
-				isActive: false,
-				isDefault: false,
-			},
 		];

 		await db.insert(models).values(modelData);
--- a/apps/chat/apps/web/src/routes/(protected)/chat/[id]/+page.svelte
+++ b/apps/chat/apps/web/src/routes/(protected)/chat/[id]/+page.svelte
@ -28,15 +28,17 @@
 	let showVersionsModal = $state(false);
 	let showDocumentPanel = $state(true);

-	// Track current request to prevent race conditions
-	let currentLoadId = $state(0);
+	// Track current request to prevent race conditions (not reactive to avoid effect loops)
+	let currentLoadId = 0;
+	let lastLoadedConversationId = '';

 	const conversationId = $derived($page.params.id ?? '');
 	const isDocumentMode = $derived(conversation?.documentMode ?? false);

 	// React to conversationId changes with race condition protection
 	$effect(() => {
-		if (conversationId) {
+		if (conversationId && conversationId !== lastLoadedConversationId) {
+			lastLoadedConversationId = conversationId;
 			loadData(conversationId);
 		}
 	});