Agent Upstart

Cool things people are building with agents — and practical help getting the most out of them.

Playbooks · Coding

Antigravity CLI 1.5.5: swarms in preview

Written November 11, 2026

Antigravity CLI 1.5.5 adds swarms in preview: a coordinator plus parallel workers on a shared message board, watched live in a full-screen board view. Google labels the feature quota-heavy itself — every worker is a separate agent that uses your quota.

The problem

One agent working through a big task does it in sequence. Splitting the work across parallel agents means each one needs its own conversation, a way to share what it finds, and a way to watch them all without losing track. Until 1.5.5, Antigravity CLI had no built-in way to run that shape of work.

What a swarm is

The 1.5.5 release notes describe it plainly: start one with /swarm <task> and the agent becomes a coordinator that adds worker agents, each in its own conversation, working on the task in parallel and collaborating through a shared message board. Workers can be forked from an existing member's conversation and can run on any model in the /model picker. When a worker's model runs out of quota, it pauses instead of retrying, and a monitor worker alerts the coordinator when the whole swarm stalls. /swarm is labelled Preview & Quota Heavy in the / menu, because every worker is a separate agent that uses your quota.

The board is the other half of the feature. /swarm board opens a full-screen live view of the current swarm — also on alt+s — with a Roster of members showing each one's state and model, a Posts feed, and a Message pane that renders the selected post as Markdown. A ● Swarm(<title>) row in the active-items tray below the prompt shows the latest post and the post count while a swarm runs, and Done when it finishes.

What else is in 1.5.5

The release before this one, 1.5.4, added Auto mode for agy acp sessions. This page is 1.5.5 only.

What to do

Try a swarm on a task that genuinely splits into parallel pieces, and watch the quota. The preview label and the quota-heavy warning are the point: a swarm multiplies agents, and each one bills against your quota, so a long swarm on a big model is the expensive way to learn whether the shape fits your work. Start small, watch the board, and check that workers pause cleanly on quota instead of burning retries. If your workflow depends on uncommitted changes surviving into a worker's copy, that fix alone is worth the update.

Limits

Related playbooks

Sources

Back to: Coding playbooks hub · Also: Antigravity CLI 1.5.4: ACP Auto mode · Antigravity CLI 1.5.3: subagent models, sandbox fix · Gemini 4 Carbon is here: $2 in, $10 out