Advertise hereReach developers building with AI coding agents. →
Gemini gemini By Emily Carter

Antigravity CLI 1.5.14 streams live thinking and plans remote work before it edits

Antigravity CLI 1.5.14 shipped on 21 November, and its headline is visibility: a new /thinking toggle streams the model's full thinking live in the transcript and keeps finished thoughts expanded without pressing Ctrl+O. The release also brings plan mode to --remote-agent sessions, so a remote agent can research and write a plan for your review before it edits files or runs commands. And the /swarm board gains a live Overview section on terminals at least 28 rows tall, showing the swarm's goal, overall progress and up to six workstream threads. Source: https://www.havoptic.com/r/antigravity-cli-1.5.14

Why it matters

The /thinking toggle fixes the most annoying part of watching a long agent task: you could see what it did, but not what it was thinking while it did it, and now the thinking streams live instead of hiding behind a keypress. Plan mode for remote agents is the safety win, because a remote agent that researches first and edits only after you approve the plan is a very different risk from one that starts editing on its own. The swarm board Overview is the same idea at team scale: goal, progress and six workstreams with status indicators, so you can see which thread is stuck without digging. The behaviour change to note is Auto mode: it now reads files outside the workspace without asking for approval, while writing or editing outside still asks. See [Antigravity on AgentStackList](https://agentstacklist.com/agents/antigravity/) for the full setup.

Who it's for

Antigravity CLI users who run long tasks or swarms, anyone driving remote agents, and Auto mode users who read files outside the workspace.

How to try it

  1. Update to 1.5.14 with your normal install route, then run /thinking on to stream the model's full thinking live in the transcript. It is saved as showThinking in settings.json and is off by default.
  2. To plan before a remote agent touches anything, start a session with --mode plan, or switch with Shift+Tab, or pick plan under Agent Mode on the Remote Agent tab of /config. The agent researches and writes a plan for your review first.
  3. Open /swarm board on a terminal at least 28 rows tall to see the live Overview: the swarm's goal, overall task progress and up to six workstream threads with status indicators for working, in review, waiting on a peer, failed or done.
  4. To pick the models behind the Orchestrator on a consumer plan, open /model and use the Orchestrator row: you can now pick both the Planner and Executor models and each one's thinking Effort.
  5. Note the Auto mode change: reading files outside the workspace no longer asks for approval, but writing or editing outside still does, and your ask and deny rules, protected files such as mcpconfig.json, and organisation policies still apply. Plan and availability: google-antigravity/antigravity-cli releases, 21 November 2026.

Source: https://www.havoptic.com/r/antigravity-cli-1.5.14

Written by Emily Carter from the original post. Checked Checked 21 Nov 2026. Features, plans and availability can change, so confirm on the source before you rely on it.

antigravitythinkingremote-agent

More Gemini tricks

All Gemini tricks →