ZeroLog
Private preview
Back to Newsroom

Product

9 October: explicit tool mode, allowlisted commands back, Compute and chat fixes

Chats use an explicit tool mode. The allowlisted-commands approval mode is available again, and Compute handles restore and previews more reliably.

User impact: Not reviewed

Added on 23 October 2026 from the repository history. Ten deploys ran on this day.

What happens to one commandThe AI asks to run a command. The chat's setting decides: with Never ask or Auto-accept edits it runs without a question; with Allowlisted commands only it runs if every part of the command is on your list; otherwise it is shown in full and waits. You can approve once, approve and switch the chat to Auto-accept edits, approve and add the pattern to your allowlist, or deny. With no answer in five minutes the request expires. A command that runs keeps its exit code and output in the chat.Never ask, Auto-accept editsAllowlisted: every part is onyour listAsk every time, or not on thelistApprove, Always in this chat,Always allowDenyno answer in 5 min1. The AI asks to run acommand2. The chat's setting decides3. Shown in full,waits for you4. The command runsexit code and outputstay in the chatDeniedthe model gets[denied by user]Request expiresthe assistant stopsand sums up
Approval modes in a chat. Diagram from the docs.
Text version

Between asking every time and accepting edits automatically, a chat can run the commands on your allowlist without asking and still ask about everything else.

New

  • Chat tool mode is explicit. Plain text conversations stay plain text.
  • Allowlisted commands mode is available in the chat settings, with a Command allowlist card in Settings.
  • Compute shows safe previews of NAS files and a verified workload lifecycle. Restoring a workspace and opening SSH access are coordinated across browser tabs.
  • Attachments and composer tips follow the workspace design. Chat tools and the workspace share the same motion.

Fixed

  • Saved messages are reconciled against the terminal history. Workspace changes are guarded, and model capacity is described more clearly.
  • Tools on the second worker work again, and worker runtime sockets keep the right ownership.

Corrections belong here too. Feed · Docs · X · Legal & privacy