Agents Machine

Chat

Conversational AI interface in the desktop app.

The Chat panel is the primary interface for interacting with AI. Messages are processed by the MCP server's router, which decides whether to respond directly, spawn an agent, or execute tools.

Features

Rich Input

Markdown support with syntax highlighting.

Tool Indicators

See when the AI calls MCP tools in real time.

Agent Status

Track which agent is running and its progress.

Code Blocks

Formatted output with copy buttons.

Message Routing

When you send a message, the route_message tool determines the best action:

Session Management

Chat sessions are automatically tracked with checkpoints stored in Qdrant.

ActionMCP ToolDescription
Search past sessionssession_historyFind conversations by content
Set contextset_contextAdd key-value context to the current session
Save checkpointsession_checkpointForce-save the current session state
View statussession_statusSee session ID, action count, buffer status

Tip: Be specific in your requests for better routing. Reference stored memories by topic for precise context.

On this page