GitHub Copilot Chat Documentation
FreeComplete chat feature guide
FreeFree tier
Inputs: text, codeOutputs: code, text
About GitHub Copilot Chat Documentation
Chat in Visual Studio Code (VS Code) enables developers to interact with AI agents using natural language. It allows asking questions about code, understanding complex logic, generating new features, fixing bugs, and more through a conversational interface. VS Code offers multiple chat surfaces: an Agents window for high-level task orchestration across projects, a Chat view in the sidebar for code assistance, Inline chat for quick in-place edits, and Quick Chat for lightweight prompts. Users can add context with #-mentions and use slash commands for common tasks. Messages can be queued or interrupt current requests with three handling options.
Key Features
Multiple chat surfaces: Agents window, Chat view, Inline chat, Quick Chat
Natural language interaction for coding tasks
Add context with #-mentions (files, symbols, etc.)
Slash commands for shortcuts and agent skills
Parallel chat sessions with background persistence
Send messages while a request is running (queue, steer, stop)
Configurable agent session: agent type, language model, permissions
Pros & Cons
Pros
- Natural language interface lowers the barrier to interacting with code
- Multiple surfaces suit different workflows (dedicated window, sidebar, inline)
- Context-aware with #-mentions for precise code references
- Non-blocking message sending allows efficient multitasking
- Session management preserves context across tasks
Best For
Add input validation to a signup formUnderstand complex code logicGenerate new featuresOrchestrate tasks across multiple projects using the Agents windowGet help with code understanding through a conversational interface
FAQ
How do I open the Chat view in VS Code?
Select the chat icon in the VS Code title bar, use the Chat: Open Chat command, or press ⌃⌘I (Windows, Linux Ctrl+Alt+I).
Can I send a new message while a chat request is running?
Yes, you have three options: Add to Queue (waits for current response to finish), Steer with Message (stops current tool execution and processes new message), or Stop and Send (cancels current request immediately). The default action is configurable.
How can I give extra context to my prompt?
You can add context by referencing files, symbols, or other information with #-mentions in your chat message.