Skip to main content
The chat interface is the primary way to interact with ScryCLI. It allows you to ask questions about your codebase, request code generation, and get AI-powered assistance.

​
Overview

The chat interface provides:
  • Natural language interaction with AI models
  • Real-time streaming responses
  • Tool execution capabilities
  • Context-aware codebase understanding

​
Using the Chat

1

Enter your prompt

Type your question or request in the input field at the bottom of the screen.
> Ask anything about your codebase...
2

Submit

Press Enter to send your message to the AI assistant.
3

View response

The AI will process your request and display the response. Responses are streamed in real-time.

​
Interface Components

​
Current Working Directory

The current working directory is displayed above the input field: 
/path/to/your/project
> Ask anything about your codebase...
You can change this using the /path command.

​
Model Indicator

The selected model is displayed in the bottom right: 
Model: openai/gpt-oss-120b:free
If no model is selected, it shows: 
Model: Not Selected
Use the /model command to select a model.

​
Response Display

Responses appear above the input area and include:
  • AI-generated text
  • Tool execution results
  • Error messages (if any)
  • Loading indicators during processing

​
Example Prompts

> Explain the authentication flow in this codebase

> What does the useChat hook do?

> Find all React components that use the SelectInput library

> Generate a new command component for /settings

​
Commands vs Chat

  • Commands (starting with /): Configuration and control operations
  • Chat prompts (regular text): AI-powered codebase assistance
When you start typing /, the command help menu appears automatically. Press Esc to dismiss it and continue with regular chat.

​
Error Handling

If an error occurs during chat:
  • The error message is displayed in the response area
  • You can retry by submitting your prompt again
  • Check that a model is selected and configured properly

​
Technical Details

The chat interface:
  • Uses the configured AI model from /model command
  • Applies a system prompt for codebase understanding
  • Supports streaming responses for real-time feedback
  • Executes tools as needed based on AI responses
  • Maintains chat context within the current session