Interface overview

Main window

  • Left — chats panel: conversations with folders, search, "+ New chat" button, token stats.
  • Center — the active chat's message feed with Markdown, code, formulas and diagrams; while the agent works — a step timeline (tools, arguments, results, tokens).
  • Top right — quick model/profile switcher, workspace folder, "Autonomous mode", "Plan mode" and "Full access" indicators.
  • Bottom — the input: text, attachments (paperclip), screenshots (Ctrl+V), templates via /, files via @, send (Enter or button).

Sidebar

At the bottom of the left panel:

  • Notes — your personal library of answer fragments.
  • AI profiles — models and providers management.
  • Servers — SSH/SFTP/FTP connections.
  • Settings — general, memory, templates, data, diagnostics, license.

Working modes

  • Autonomous mode (chat header) — enables the agent loop: the model can call tools (files, commands) to complete a task.
  • Plan mode — the agent agrees on a plan with you before acting.
  • Full access — allows local operations without step-by-step confirmations (remote ones are always confirmed). Dangerous operations still require explicitly enabling this toggle.
← Docs