First Run

Launch Gravity

On first launch, you'll see the splash screen followed by the main window with the chat interface. The application initializes its agent pool, loads dynamic agents from agents/*.json, and establishes the MCP server connections configured in settings.json.

Configure Provider

1. Click the Settings icon in the ribbon bar (or press Ctrl+Comma)
2. Select your LLM provider from the dropdown:
   • Ollama - Local models, default http://localhost:11434
   • LM Studio - Local server, default http://localhost:1234
   • OpenRouter - Cloud gateway, https://openrouter.ai/api/v1
   • OpenAI - https://api.openai.com/v1
   • Groq - Fast inference, https://api.groq.com/openai/v1
   • Google AI Studio - https://generativelanguage.googleapis.com/v1beta/openai/
   • TogetherAI - https://api.together.xyz/v1
   • GitHub Models - https://models.inference.ai.azure.com
3. Enter the base URL and API key (if required for cloud providers)
4. Choose a model (e.g., qwen2.5-coder:1.5b for Ollama, gpt-4o for OpenAI)
5. Click Test Connection to verify the provider is reachable
6. Save settings - the app will validate the model on next task

Run Your First Task

Type a natural language request in the chat input, for example:

"Read Program.cs and explain the dependency injection setup"

The agent will: 1. Use code_editor.read_file to read the file 2. Analyze the code using its reasoning capabilities 3. Respond with an explanation in the chat Try other starter tasks:

• "List all files in the Controllers folder"
• "Create a new Razor view for a contact form"
• "Run dotnet build and show me any errors"