Gemini CLI in NeuroSquadGemini CLI,with a canvas around it
Run as many Gemini CLI agents as you need, each in its own card with a real terminal. NeuroSquad knows when each one is working, waiting on you or finished, hands it terminals, browsers and notes by arrow — and the new tools arrive while it runs.
Official sitegeminicli.com- The real
geminiCLI, your own login - Nothing written to your
~/.gemini - Free, for Windows 10 and 11
It knows when Gemini is done — and when it’s asking
A quiet terminal can mean “finished” or “waiting for your confirmation”. NeuroSquad doesn’t guess: every Gemini CLI card reports its own turns through Gemini CLI’s own hooks.
BeforeAgenthookYou press Enter and Gemini fires BeforeAgent. The card’s icon spins, the Squad status card counts it as working, and the toast from its last question is taken down.
NotificationhookGemini opens a confirmation and sends a ToolPermission notification with the details — the command, the file or the MCP tool. The card pulses, and the toast leads with exactly that.
AfterAgenthookAfterAgent closes the turn — a cancelled one too. The next queued prompt goes out, the journal gets the answer, and a toast says it’s done.
Its questions too
When Gemini asks you something with ask_user, the question itself comes to the card and the toast: “Gemini CLI is asking: Tabs or spaces?”.
One status everywhere
The card, the sidebar, the Squad status card, custom cards and the phone all read the same status.
Your hooks stay yours
NeuroSquad’s hooks are added to the card’s own settings layer; the hooks in your settings keep running.
Three things you’ll do every day
Replicas of the app, played step by step. When Gemini asks for confirmation, the answer is yours.
Gemini CLI fixes a failing test through a connected terminal, stops to ask before deleting a folder, and finishes — the journal writes itself.
Everything Gemini CLI can do in NeuroSquad
Gemini CLI is integrated as fully as Claude Code. Here is all of it, with a docs page for each — and, at the end, what it honestly doesn’t get.
- 01
Knows its state
Gemini CLI’s own hooks report every turn, so nothing is guessed.
Finished and waiting are different
Each card gets Gemini CLI’s own lifecycle hooks:
Docs: Finished and waiting are differentBeforeAgentmarks it working, aToolPermissionNotification— waiting on you,AfterAgent— finished.SessionStartfollows/clearto the new session.A notification that says what it needs
The toast leads with what Gemini wants to do —
Docs: A notification that says what it needsConfirm Shell Command — rm -rf dist, the file of an edit, or the MCP tool. One per agent; a waiting one stays until you answer.Every card is its own session
The card’s id is the session id:
Docs: Every card is its own session--session-idon the first launch,--resumeafter, so restarting the app continues the conversation. A session that was never used starts fresh by itself.The Squad status card
Every AI agent of the workspace on one card, with its status, the live time of its turn and its model — from Gemini’s own log, or your default (“auto”) before the first turn.
Docs: The Squad status cardCustom cards can follow it
Cards built with the Card SDK see the same statuses, turn events and last replies, and
Docs: Custom cards can follow itagent_waitgets a Gemini helper’s answer.
- 02
Works through the canvas
An arrow from the card is a set of tools. Draw it, and Gemini can use what is at the other end — right away.
Arrows are tools
An arrow to a browser, terminal, note or another agent gives Gemini
Docs: Arrows are toolsbrowser_*,terminal_*,note_*oragent_*over NeuroSquad’s MCP server, pre-trusted. Each call lights up its arrow.New arrows work mid-session
Gemini CLI re-reads the tool list when it changes, so an arrow drawn while it works adds the tools to its very next request. No restart, no new session.
Docs: New arrows work mid-sessionIt can add cards itself
Docs: It can add cards itselfcanvas_spawn_cardlets it open a terminal, a browser, a note or a helper agent — Gemini or any other CLI — next to itself, already joined by an arrow.Canvas mode
Gemini’s own shell and web tools are excluded (
Docs: Canvas modetools.exclude), so commands run in terminal cards you can watch. Excluded tools never reach the model — even with--yolo.Skills from skills.sh
An arrow to a skill card gives Gemini
Docs: Skills from skills.shskill_loadwith each skill and when to use it — and like every arrow, it works without a restart.
- 03
Built for long runs
Hours-long sessions run into full contexts and quotas. Both are handled on the card.
Context meter and Compress
A small ring in the header shows how full the context is, against Gemini’s own window for the model. Compact sends
Docs: Context meter and Compress/compressfor you.Continues after a quota
Gemini retries short rate limits itself. When it hits a daily quota and shows its own “Usage limit reached” dialog, the card reads the reset time, waits and sends “continue” then.
Docs: Continues after a quotaPrompt queue
Line up the next messages: each goes out on
Docs: Prompt queueAfterAgent, when the turn is really over — never in a pause mid-thought. Daily prompts too.Hand off to a fresh agent
When a session is full, start a new agent beside it — Gemini or any other CLI — with a summary of the work typed in, ready for you to send.
Docs: Hand off to a fresh agentJournal
Each finished answer is added to a connected note, read from Gemini’s own chat log: a running record of what the agent did.
Docs: Journal
- 04
Works in a team
Several Gemini agents — or a mix of CLIs — on one project, without touching your own Gemini setup.
Your
~/.geministays untouchedEach card runs with a home of its own (
Docs: Your ~/.gemini stays untouchedGEMINI_CLI_HOME): its settings layered over yours, everything else linked. Your login, history and sessions are shared; nothing of yours is written.Isolated worktrees
Give an agent its own git worktree, so parallel agents never edit the same files. A setup command and files to copy, like
Docs: Isolated worktrees.env, prepare it before Gemini starts.Add a reviewer
One click starts an agent on a different CLI in the same worktree, with an arrow from the author and a prompt to review without editing anything.
Docs: Add a reviewerDangerous mode, per card
For a run you trust, a card can start Gemini with
Docs: Dangerous mode, per card--yolo. Itsask_userquestions still come to you. A per-card switch, off by default.One instructions file
You edit AGENTS.md; NeuroSquad mirrors it to GEMINI.md and never overwrites a changed file without showing you the diff.
Docs: One instructions file
- 05
Cost, control and access
What it used, what it may spend, which model it runs on, and how you reach it.
Tokens per card
Every request’s tokens — input, cached, output, thoughts — read from Gemini’s own chat logs, subagents folded into their session, split exactly by card.
Docs: Tokens per cardBudget brake
A limit per workspace. Past it, each agent’s turn is cancelled — the session is kept — and automatic prompts stop. Raise the limit and it’s lifted.
Docs: Budget brakeOpenRouter models
Run a card on any OpenRouter model. Gemini speaks only the Gemini API, so the card talks to a small gateway inside NeuroSquad; your OpenRouter key never reaches Gemini.
Docs: OpenRouter modelsMCP servers from the registry
Install a server from the official MCP Registry once, then give it to an agent with an arrow. It isn’t pre-trusted: Gemini asks before every call to it.
Docs: MCP servers from the registryFrom your phone
Scan a QR code and the whole canvas opens in the phone’s browser: read Gemini’s terminal, answer its confirmation, send the next prompt.
Docs: From your phone
- 06
Not available for this CLI
What Gemini CLI does not expose, or the app does differently for it. Said plainly.
Side calls aren’t counted
Gemini’s calls outside the chat — the model router’s classifier, session summaries — are only in its telemetry, not its logs. Usage shows the chat’s requests, exactly.
Docs: Side calls aren’t countedNo price for Google’s own models
There are no Gemini list prices in NeuroSquad yet, so cards on your Google login show tokens and “no price” — never a made-up $0. OpenRouter cards are priced from its catalogue.
Docs: No price for Google’s own modelsHooks cost a moment
Gemini CLI’s hooks are processes it waits for — about a quarter of a second each on Windows. That’s why only the turn-level events are used, not every tool call.
Docs: Hooks cost a momentGemini-only features stay on Google
Through the OpenRouter gateway, what OpenAI-style chat has no equivalent for — Google Search grounding, code execution, thought signatures — does not survive.
Docs: Gemini-only features stay on Google
Who’s waiting, and what it used
Two cards that answer the questions you ask most when several Gemini agents run at once.
The Squad status card
Every agent of the workspace with its status, model and turn time. The one waiting on you comes first, with what it asks. Tap a status to filter.
- coupon-fixGemini CLIgemini-2.5-proGemini CLI needs your permission: Confirm Shell Command — rm -rf distNeeds you2:34
- api-migrationGemini CLIgemini-2.5-proMove routes to the new clientWorking11:07
- lib-researchGemini CLIautoCompare two date librariesWorking5:25
- docs-passGemini CLIgemini-2.5-flashUpdate the READMEFinished3 min. ago
Usage & cost
Read from Gemini CLI’s own chat logs on your disk and split by card. The rows always add up to the total.
| Agent | Tokens | Cost | ||
|---|---|---|---|---|
| coupon-fix | 912,406 | $2.14 | ||
| api-migration | 604,118 | $1.39 | ||
| docs-pass | 188,530 | $0.09 | ||
| Total | 79 | 1,705,054 | $3.62 |
The real Gemini CLI, plus a home of its own
NeuroSquad starts the same Gemini CLI you run yourself, in a real terminal in the card. This is everything it adds:
Nothing written to your ~/.gemini
Settings made inside a card are kept in NeuroSquad’s own folder and never written back. Deleting a card unlinks before it removes.
Your login, your plan
Gemini CLI runs with the Google login or API key it already uses. The MCP token is expanded by Gemini from the environment — never stored in a file.
A session is your session
Because the card’s home links your real sessions, gemini --resume <id> works from a plain terminal too — and the other way round.
Gemini CLI in NeuroSquad, answered
Give Gemini CLI a canvas
Free for Windows 10 and 11. Your code, your login and your usage stay on your machine.