Qwen Code in NeuroSquadQwen Code,with a canvas around it
Every card runs the real qwen in its own terminal, with your own login or key — or on OpenRouter. Qwen’s hooks tell NeuroSquad when each one works, asks or finishes, arrows give it terminals, browsers and other agents, and its usage is counted request by request.
Official siteqwenlm.github.io/qwen-code-docs- The real
qwen, your own login or key - Nothing written to
~/.qwen - Free, for Windows 10 and 11
It knows when Qwen is done — and when it’s asking
Qwen Code has Claude Code’s hook system, with an http hook type. Each card’s settings post every event straight to NeuroSquad — no process per hook, nothing guessed.
UserPromptSubmithookYou press Enter and Qwen fires UserPromptSubmit. The card’s icon spins, the Squad status card counts it as working, and the toast from its last question is taken down.
PermissionRequesthookPermissionRequest names the tool and its input — “run_shell_command — rm -rf dist” — or carries Qwen’s question word for word. The card pulses and the toast leads with it until you answer.
StophookStop ends the turn and brings the context in use along. The next queued prompt goes out, the journal gets the answer, and a toast says it’s done.
Failures end a turn too
An API error arrives as StopFailure: the card finishes, and a rate limit or billing error shows on the card.
Subagents don’t confuse it
Events from inside a subagent don’t move the card’s status — but a subagent’s permission request still asks you.
Click, and you’re there
A toast click switches to the workspace, flies the camera to the card and flashes it — even among dozens of agents.
Three things you’ll do every day
Replicas of the app, played step by step. When Qwen asks for permission, the answer is yours.
Qwen Code fixes a failing test, asks before deleting a folder, and finishes — the journal writes itself.
Everything Qwen Code can do in NeuroSquad
Qwen Code gets NeuroSquad’s full treatment through one per-card settings file. One thing it can’t do is marked below.
- 01
Knows its state
Qwen’s own hooks report every turn, so nothing is guessed.
Finished and waiting are different
Each card’s settings add Qwen
Docs: Finished and waiting are differenthttphooks:UserPromptSubmitmarks it working,PermissionRequest— waiting on you,Stop— finished.A notification that says what it needs
The toast leads with Qwen’s own words —
Docs: A notification that says what it needsrun_shell_command — rm -rf dist, or the question it asks you. One per agent; a waiting one stays until you answer.Every card is its own session
Docs: Every card is its own session--session-idwith the card’s id on the first launch,--resumeafter./clearstarts a new session in the same process — its id comes with the hooks, so the next restart resumes the right one.The Squad status card
Every agent of the workspace with its status, its model from Qwen’s own usage log (or your default before the first turn) and the live time of its turn.
Docs: The Squad status card- Emulated
Cancelled turns end too
Esc and Ctrl+C fire no
Docs: Cancelled turns end tooStop. NeuroSquad closes the turn when Qwen reports it is idle again (Notificationidle_prompt), so the card doesn’t hang on “working”.
- 02
Works through the canvas
An arrow from the card is a set of tools. Draw it, and Qwen can use what is at the other end.
- Not possible
No live tool updates
Arrows are tools as everywhere —
Docs: No live tool updatesterminal_*,browser_*,note_*,agent_*— but Qwen can’t pick up a changed tool list mid-session. An arrow drawn while it runs takes effect on its next start. It can add cards itself
Docs: It can add cards itselfcanvas_spawn_cardlets it open a terminal, a browser, a note or a helper agent next to itself, already joined by an arrow.Canvas mode
Docs: Canvas moderun_shell_command,monitor,web_fetchandweb_searchare disabled and denied in the card’s settings. That isn’t an approval, so--yolodoesn’t bring them back.MCP servers from the registry
Install a server from the official MCP Registry once, then give it to an agent with an arrow. Qwen asks before every call to it; NeuroSquad’s own tools load up front instead of hiding behind Qwen’s tool search.
Docs: MCP servers from the registrySkills from skills.sh
An arrow to a skill card gives Qwen
Docs: Skills from skills.shskill_loadwith each skill and when to use it (from its next start, like every new arrow).
- 03
Built for long runs
Full contexts and usage limits, both handled on the card.
Context meter and Compact
Every
Docs: Context meter and CompactStopbrings the tokens in use and the window Qwen compresses against, so the ring in the header is exact. Compact sends/compress.Continues after a usage limit
A
Docs: Continues after a usage limitStopFailurefor a rate limit or billing error marks the card, and when the provider says when it resets, “continue” goes out then.Prompt queue
Line up the next messages: each goes out on
Docs: Prompt queueStop, never on a pause mid-thought. Daily prompts too.Hand off to a fresh agent
When a session is full, start a new agent beside it — Qwen or any other CLI — with a summary and Qwen’s last answer typed in, ready for you to send.
Docs: Hand off to a fresh agentJournal
Each finished answer, read from Qwen’s own chat log, is added to a connected note: a running log of what the agent did.
Docs: Journal
- 04
Works in a team
Several Qwen agents — or a mix of CLIs — on one project without stepping on each other.
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 first.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 starts Qwen with
Docs: Dangerous mode, per card--yolo. Off by default; canvas mode still holds.Roles
Reviewer, tester, researcher, architect, writer or your own: the role is typed in as a first prompt you can see and edit.
Docs: RolesOne instructions file
Qwen reads AGENTS.md as well as QWEN.md, so the file your other agents use works for Qwen as is — and NeuroSquad can mirror it to QWEN.md too.
Docs: One instructions file
- 05
Cost, control and access
What it spends, what it may spend, which model it runs on, and how you reach it.
Exact usage and cost
Qwen logs every request, subagents and background calls included. NeuroSquad reads that log per card, to the token; OpenRouter cards are priced from OpenRouter’s catalogue.
Docs: Exact usage and costBudget brake
A limit per workspace. Past it, Qwen gets Ctrl+C — its turn is cancelled, Qwen stays open — and automatic prompts stop until you raise the limit.
Docs: Budget brakeOpenRouter models
Run a card on an OpenRouter model instead of your own login. The key goes only into that card’s environment, never into arguments or logs.
Docs: OpenRouter modelsDictation into the prompt
Hold a hotkey and speak. Speech is recognised on your machine and typed into the active agent’s prompt.
Docs: Dictation into the promptFrom your phone
Scan a QR code and the whole canvas opens in the phone’s browser: read Qwen’s terminal, answer its question, send the next prompt.
Docs: From your phone
Who’s waiting, and what it cost
Two cards that answer the questions you ask most when several Qwen Code sessions 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-fixQwen Codeqwen/qwen3-coderQwen Code needs your permission: run_shell_command — rm -rf distNeeds you2:34
- api-migrationQwen Codeqwen/qwen3-coderMove routes to the new clientWorking12:24
- search-uiQwen Codeqwen3-coder-plusRework the search boxWorking5:25
- docs-passQwen Codeqwen/qwen3-coderUpdate the READMEFinished3 min. ago
Usage & cost
Read from Qwen’s own request log on your disk and matched to the card by session, /clear included. The totals match Qwen’s log to the token.
| Agent | Tokens | Cost | ||
|---|---|---|---|---|
| coupon-fix | 1,532,904 | $0.58 | ||
| api-migration | 906,118 | $0.34 | ||
| docs-pass | 241,560 | no price | ||
| Total | 130 | 2,680,582 | $0.92 |
The real Qwen Code, plus one settings file
NeuroSquad starts the Qwen Code you installed, in a real terminal in the card. Almost everything travels in one per-card settings file Qwen merges over yours:
Nothing written to ~/.qwen
The settings file is per card in NeuroSquad’s own data folder, and your hooks and deny rules are kept — Qwen joins the lists.
No secret on disk
The per-card token is written as ${NEUROSQUAD_MCP_TOKEN}; Qwen expands it from the card’s environment.
Your administrator’s rules win
A machine-wide Qwen settings file is merged in with its values on top.
Qwen Code in NeuroSquad, answered
Give Qwen Code a canvas
Free for Windows 10 and 11. Your code, your keys and your costs stay on your machine.