Skip to content
Cheffed

Chat Interface

The chat interface is where you interact with your AI agent. Understanding its features helps you work more effectively.

Interface Layout

Section titled “Interface Layout”

The chat interface consists of:

  • Message History - Scrollable conversation with the agent
  • Input Area - Where you type messages and commands
  • Session Info - Header showing session details and status

Sending Messages

Section titled “Sending Messages”

Type your message in the input area and press Enter or click Send. You can:

  • Ask questions about code
  • Request changes or new features
  • Share error messages for debugging
  • Provide feedback on agent outputs

Multi-line Messages

Section titled “Multi-line Messages”

For longer messages, use Shift+Enter for new lines within your message.

Agent Responses

Section titled “Agent Responses”

Agent responses can include:

Text Responses

Section titled “Text Responses”

Explanations, analysis, and conversational replies appear as formatted text with syntax highlighting for code snippets.

Tool Calls

Section titled “Tool Calls”

When the agent takes action, you’ll see tool call visualizations:

Bash Commands

Section titled “Bash Commands”
$ npm install lodash

Shows the command executed and its output.

File Operations

Section titled “File Operations”

Read - Displays the file being read with syntax highlighting

Write - Shows the new file content being created

Edit - Highlights the specific changes being made to existing files

Search Operations

Section titled “Search Operations”

Glob - Shows file patterns being searched

Grep - Displays search patterns and matching results

Structured Follow-ups

Section titled “Structured Follow-ups”

Sometimes the agent offers structured options for next steps. Click a suggested action to send it as your next message.

Real-time Updates

Section titled “Real-time Updates”

The interface updates in real-time as the agent works:

  • See commands as they execute
  • Watch files being read and written
  • Monitor long-running operations
  • Receive streaming text responses

Session Artifacts

Section titled “Session Artifacts”

Agents can produce artifacts during a session:

  • Code files - New or modified source files
  • Documentation - Generated README or docs
  • Reports - Analysis summaries or audit reports

Access artifacts from the session header menu.

Conversation Tips

Section titled “Conversation Tips”

Be Specific

Section titled “Be Specific”

Instead of:

“Fix the bug”

Try:

“Fix the TypeError on line 45 of utils.ts - the user object can be null but we’re accessing .name without checking”

Provide Context

Section titled “Provide Context”

Share relevant information:

  • Error messages and stack traces
  • Expected vs. actual behavior
  • Related files or dependencies

Iterate

Section titled “Iterate”

Use follow-up messages to refine:

  • “That’s close, but also handle the edge case where…”
  • “Good, now add error handling for…”
  • “Can you explain why you chose this approach?”

Review Before Applying

Section titled “Review Before Applying”

Always review changes before the agent applies them:

  • Read the proposed code changes
  • Understand the approach
  • Ask questions if something seems wrong

Keyboard Shortcuts

Section titled “Keyboard Shortcuts”
ShortcutAction
EnterSend message
Shift+EnterNew line in message
EscapeClear input

Next Steps

Section titled “Next Steps”