Configuration

After installing Astonish, you need to configure it to work with an AI provider. This guide will walk you through the configuration process.

Configuration Files

Astonish uses two main configuration files:

1. config.ini: General configuration settings
2. mcp_config.json: MCP (Model Context Protocol) server configuration

These files are automatically created and managed by the application in the user's config directory:

• On Linux: ~/.config/astonish/
• On macOS: ~/Library/Application\ Support/astonish/
• On Windows: %APPDATA%\astonish\

Setting Up an AI Provider

Astonish supports multiple AI providers. To set up a provider, use the setup command:

astonish setup

This will start an interactive setup process:

1. Astonish will display a list of available providers
2. Select a provider by entering its number
3. Follow the provider-specific configuration steps

Supported Providers

Astonish currently supports the following AI providers:

Provider	Description
Anthropic	Access to Claude models
LM Studio	Integration with locally running LM Studio
Ollama	Integration with locally running Ollama models
Openrouter	Access to various models through Openrouter
SAP AI Core	Integration with SAP AI Core

Provider-Specific Configuration

Each provider has different configuration requirements:

Anthropic

• API Key: Required for authentication
• Model: The Claude model to use (e.g., claude-3-opus-20240229)

LM Studio

• Host: The host where LM Studio is running (default: localhost)
• Port: The port LM Studio is listening on (default: 1234)
• Model: The model name (optional)

Ollama

• Host: The host where Ollama is running (default: localhost)
• Port: The port Ollama is listening on (default: 11434)
• Model: The model name (e.g., llama2)

Openrouter

• API Key: Required for authentication
• Model: The model to use (e.g., openai/gpt-4)

SAP AI Core

• API Key: Required for authentication
• URL: The SAP AI Core API URL
• Deployment ID: The deployment ID for the model

MCP Configuration

MCP (Model Context Protocol) allows Astonish to extend its capabilities by connecting to external tools and resources. To configure MCP servers:

astonish tools edit

This will open the MCP configuration file in your default editor. The configuration is a JSON file with the following structure:

{
  "mcpServers": {
    "server-name": {
      "command": "command-to-run-server",
      "args": ["arg1", "arg2"],
      "env": {
        "ENV_VAR1": "value1",
        "ENV_VAR2": "value2"
      },
      "transport": "stdio"
    }
  }
}

Example MCP Configuration

Here's an example configuration for a weather MCP server:

{
  "mcpServers": {
    "weather": {
      "command": "node",
      "args": ["/path/to/weather-server/index.js"],
      "env": {
        "OPENWEATHER_API_KEY": "your-api-key"
      },
      "transport": "stdio"
    }
  }
}

Advanced Configuration

Logging

Astonish logs information to help with debugging and monitoring. You can control the verbosity of logging using the --verbose flag.

Next Steps

After configuring Astonish, you're ready to:

1. Try the Quick Start Guide to create your first agent
2. Learn about Agentic Flows to understand how Astonish works
3. Explore the Commands Reference for more details on available commands