IBM i MCP Server

# Configuration Guide Centralized, type-safe configuration for the IBM i Agent Infrastructure. The system uses two configuration files: 1. **`.env`** - API keys, database credentials, MCP connection (sensitive data) 2. **`config.yaml`** - Agent behavior, model selection, UI settings (version controlled) --- ## Configuration Files ### 1. Environment Variables (`.env`) Create `infra/.env` with your API keys and connection settings: ```bash cp infra/.env.example infra/.env ``` ```bash # ============================================================ # MCP Server (Required) - IBM i database access # ============================================================ # The MCP server runs automatically in Docker Compose # Use ibmi-mcp-server:3010 for container-to-container communication MCP_URL=http://ibmi-mcp-server:3010/mcp MCP_TRANSPORT=streamable-http # ============================================================ # AI Model Provider (Choose at least one) # ============================================================ # Option 1: watsonx (IBM Cloud) # Get keys from: https://cloud.ibm.com WATSONX_API_KEY=your_ibm_cloud_api_key WATSONX_PROJECT_ID=your_project_id WATSONX_URL=https://us-south.ml.cloud.ibm.com WATSONX_MODEL_ID=meta-llama/llama-3-3-70b-instruct # Option 2: OpenAI # Get key from: https://platform.openai.com/api-keys OPENAI_API_KEY=sk-your_openai_key # Option 3: Anthropic # Get key from: https://console.anthropic.com ANTHROPIC_API_KEY=sk-your_anthropic_key ``` > **Note**: The default model is `"anthropic:claude-haiku-4-5"` #### Getting API Keys **watsonx (IBM Cloud):** 1. Sign up at [cloud.ibm.com](https://cloud.ibm.com) 2. Navigate to IBM watsonx.ai 3. Create a project and note the Project ID 4. Generate an API key from IAM (Identity & Access Management) **OpenAI:** 1. Sign up at [platform.openai.com](https://platform.openai.com) 2. Navigate to API Keys 3. Click "Create new secret key" 4. Copy the key immediately (shown only once) **Anthropic:** 1. Sign up at [console.anthropic.com](https://console.anthropic.com) 2. Navigate to API Keys 3. Create a new API key 4. Copy the key immediately (shown only once) #### MCP Server Configuration The MCP server runs automatically as part of the Docker Compose stack: - **Container name**: `ibmi-mcp-server` - **Port**: `3010` - **Health endpoint**: `http://localhost:3010/health` - **MCP endpoint**: `http://ibmi-mcp-server:3010/mcp` (inside Docker network) **Network Configuration:** - **Inside Docker network**: Use `http://ibmi-mcp-server:3010/mcp` (recommended) - **From host machine**: Use `http://localhost:3010/mcp` - **From external**: Use `http://host.docker.internal:3010/mcp` The MCP server configuration is loaded from the monorepo root `.env` file (not `infra/.env`). --- ### 2. Agent Configuration (`config.yaml`) The `config.yaml` file controls agent behavior, model selection, and UI settings. #### Agent Configuration **performance monitoring agent example:** ```yaml agents: # Default model for all agents (can be overridden per agent) default_model: "anthropic:claude-haiku-4-5" # Performance monitoring agent ibmi-performance-monitor: # Uses default_model when not specified # model: enable_reasoning: false debug_mode: false ``` Override the default model per agent as needed. Supported models include: - **watsonx**: `meta-llama/llama-3-3-70b-instruct`, etc. - **OpenAI**: `gpt-4o`, etc. - **Anthropic**: `claude-sonnet-4-5-20250929`, etc.