Skip to content

The /newtask Command

/newtask is a built-in slash command that duplicates the current conversation into a new task. Use it when you want to branch off — try a different direction, or keep working on a follow-up — without losing or disturbing the original chat.

When you send /newtask in the chat input:

  1. The command is handled directly in the request loop — no model call is made, and the literal /newtask text is not sent as a chat turn.
  2. Kodik creates a new task named Copy of <original title> and selects it.
  3. The full visible transcript and the full provider conversation history are carried over verbatim — the real turns, not a summary. The model in the new task knows exactly what happened, word for word.
  4. The original task is left untouched; the two tasks diverge from that point on.

Picking /newtask from the slash-command menu inserts it into the input rather than running it immediately.

If the conversation is empty, Kodik shows a notification saying there is nothing to copy and no new task is created.

You’ve explored a problem and want to try two different implementations without mixing them up: send /newtask, and continue one approach in the copy while the original stays as it was.

Unlike /compact, /newtask does not shrink the context — it duplicates it. If the conversation is close to the context limit, run /compact in the copy (or before duplicating) to free up space.