Skip to main content

​
Find your way around

Tomdacat Code v3.3 with its icon rail, model selector, composer modes, and inline effort slider A slim icon rail on the left is always visible: Chat, History, Subagents, and Settings at the bottom. You’re on Chat by default — the other three are one click away instead of hidden behind a header icon.

​
Send your first message

Type into the input bar at the bottom. It auto-grows as you type — up to about five lines — before an internal scrollbar takes over. Press + to reveal the Normal/Plan/Auto mode selector, image attachment, and the inline reasoning effort slider together. The plus springs into a warm-colored ×; press it again to collapse the controls. Reasoning controls appear only for supported models and need no second menu. For more room for thinking and the answer, choose a token limit in Settings → Generation. The limit is per model request; it does not increase your plan allowance.
Press Enter to send (Shift+Enter inserts a newline instead). You’ll see:
  1. A thinking indicator while the request is in flight.
  2. The assistant’s reply streaming in live, word by word.
  3. If the agent needs to call a tool (like read_tree here), a tool card appears and fills in with the call’s arguments as they stream — see Chat & Streaming for the full picture.

​
Approve your first tool call

Read-only tools like read_tree run automatically. The moment the agent tries something that changes your place — editing a script, creating an instance, anything tagged mutating — it stops and asks: A tool call card filling in live while streaming, then resolving into Deny / Allow / Always Allow buttons
  • Deny — the agent gets back "Denied by user." and adapts.
  • Allow — runs once, asks again next time.
  • Always Allow — runs now and silently for every future call to that tool, until you change it back in Settings.
If the tool is edit_script, a second dock widget pops open with a real side-by-side diff so you can see exactly what’s about to change before you decide.

​
Useful slash commands

Type / in the input box to see the full list with autocomplete. The most common ones: See Mentions & Commands for the full reference, including @ mentions for instances, scripts, and skills.

​
What’s next

Mentions & Commands

Pull instances, scripts, and skills into a message with @, plus every slash command.

Tools & Permissions

See every tool in the catalog and how the permission gate works.

Models & Reasoning

Pick a model, read its badges, and tune reasoning effort.