Stardew-Sage

Stardew Sage is an MCP-enabled AI assistant for Stardew Valley that blends a conversational agent with a rich backend data layer (MySQL) and a Redis-backed vector store for semantic search. The MCP framework enables context-aware responses, RAG-style Q&A, and multimodal analysis (image uploads). After starting the server, you can access the frontend interface for chat, RAG queries, image analysis, and data management. The MCP API endpoints exposed by the backend allow you to stream birthday stories, daily missions, and detailed database queries via Server-Sent Events (SSE). Use the frontend to interact with the agent or call the MCP endpoints directly for programmatic access.

Prerequisites:

• Docker and Docker Compose (optional if you deploy with a single container)
• MySQL 8.0+
• RedisStack (with RediSearch)
• Java 17+ (for local build) or use the provided Docker image
• AI API key for OpenAI/GLM-compatible providers

1. Prepare the database

• Install MySQL 8.0+
• Create and seed the database schema from database.sql (provided in the repo) or import ry-vue schema.

2. Install and run RedisStack

• Option A (Docker): docker run -d --name redis-stack -p 6379:6379 -p 8001:8001 redis/redis-stack:latest
• Confirm RedisStack is running: redis-cli PING -> PONG

3. Build or pull Stardew Sage MCP server image

• If using Docker image (recommended): docker pull stardew-sage:latest

• If building locally (optional): mvn clean package -DskipTests docker build -t stardew-sage:latest .

4. Configure the MCP server

• Set the following environment variables (example): SPRING_DATASOURCE_URL=jdbc:mysql://localhost:3306/ry-vue SPRING_DATASOURCE_USERNAME=root SPRING_DATASOURCE_PASSWORD=your_password SPRING_REDIS_HOST=localhost SPRING_REDIS_PORT=6379 AI_OPENAI_API_KEY=your_openai_api_key AI_VECTORSTORE_INDEX=stardew_index AI_BASE_URL=https://api.openai.com/v1 DATABASE_NAME=ry-vue

5. Run the MCP server (via Docker)

• If using the docker run command from mcp_config, execute: (Ensure ports 8099, 8080, and 80 are available) docker run -i --name stardew-sage -p 8099:8099 -p 8080:8080 -p 80:80
  -e SPRING_DATASOURCE_URL=jdbc:mysql://localhost:3306/ry-vue
  -e SPRING_DATASOURCE_USERNAME=root
  -e SPRING_DATASOURCE_PASSWORD=your_password
  -e SPRING_REDIS_HOST=localhost
  -e SPRING_REDIS_PORT=6379
  -e AI_OPENAI_API_KEY=your_openai_api_key
  -e AI_VECTORSTORE_INDEX=stardew_index
  -e AI_BASE_URL=https://api.openai.com/v1
  stardew-sage:latest

6. Access the application

• Frontend/UI: http://localhost:80
• MCP API: http://localhost:8099
• Admin API: http://localhost:8080

Note: If you run into port conflicts or environment variable issues, adjust the docker run command accordingly and ensure MySQL and RedisStack are accessible from inside the container.

Tips and common issues:

• Ensure RedisStack includes RediSearch support; otherwise vector-based search will fail.
• The MCP integration relies on an OpenAI-compatible API; confirm API keys and base URLs are correctly set.
• If the frontend fails to load, verify the frontend build assets are available and that the UI is configured to communicate with the MCP endpoint at /api/.
• For local development, you can run aurora-mcp and aurora-ui separately; the README exposes port mappings (8099 for MCP, 8080 for Admin, 80 for frontend).
• Review application.yml defaults to customize database connections, Redis, and AI providers per environment.
• If you update the database schema, re-run migrations or ensure your MyBatis mappings align with the current schema.