aisquare
One terminal over your projects and your coding agents: task a manager in plain words, and it runs the coders, testers and reviewers for you.
Type asq and you get a full-screen, mouse-driven view: your projects on the
left, and on the right a manager agent you task in prose — it plans, spawns
coders, testers and reviewers, and loops until the goal is met. Click any of
them and you are inside its real Claude Code session, typing at it directly.
Nothing is relayed or re-rendered as a chat.
Underneath, agents get a memory that persists across sessions — your preferences, each project's conventions — so every session starts oriented instead of cold. The memory also works on its own, without the UI.
Quickstart 1: memory only
Needs uv and
Claude Code; no installer, no tmux. Node 22+
is optional and adds the codebase snapshot. Run init inside your repo:
UV_PYTHON_DOWNLOADS=automatic uv tool install --python 3.13 aisquare-cli
cd path/to/your/repo
aisquare init --local --agent claude-code
aisquare remember --user "prefer pytest over unittest"
init registers the repo you run it in and wires Claude Code's hooks, so every
new session starts with what you told it to remember. aisquare context list
shows what is in scope;
Memory
has the rest. To install from inside Claude Code instead, see
the Claude Code plugin.
Quickstart 2: the full fleet
One line. It works out what your machine already has, installs only what is missing, and ends by offering to open the UI:
curl -fsSL https://raw.githubusercontent.com/AISquare-Studio/aisquare-cli/main/install.sh | sh
macOS, Linux and WSL2. It installs uv, a Python 3.13 for the CLI alone,
aisquare-cli, tmux, gh, git, Node and Claude Code, then registers the git repo
you ran it from and wires Claude Code's hooks. Running it again is a no-op.
Install
has the PowerShell line that sets up WSL2 on Windows, how to read the script
before you run it, and how to install by hand.
Then asq opens on Welcome, three steps to a running fleet:
- Project: use the folder you are in, or choose another.
- Claude Code: connect it. If it is missing, Welcome names the install command; if it is signed out, it offers the sign-in.
- Fleet: press Start manager, then Open the manager and answer Claude Code's one-time question about trusting the folder, then Start the coders: a manager and two coders, each a real Claude Code session. Type your goal to the manager in prose.
Click any agent to see and drive its session. Press F12 to hand focus
back to the sidebar, where w returns to Welcome and q quits. The agents
keep running; reopen asq and it re-attaches to what it finds.
The manager never writes code and never merges — a human does that.
Everything is also a command
Everything the UI does is also a plain command, and every one takes --json:
aisquare fleet ls # this project's agents and their live state
aisquare fleet attach # the same session in raw tmux, full fidelity
aisquare doctor # is everything wired? (and how to fix anything)
What stays free
Everything that runs on your machine — memory, the fleet, the board — is MIT
licensed, needs no account, and stays free. It's a single CLI, local-first,
backed by one SQLite file: no daemon, no cloud dependency. Sign in with
aisquare login only when a command needs to act as you on AISquare.
Requirements
- macOS, Linux, or Windows through WSL2.
- Python 3.11+ if you install by hand (the one-liner brings its own 3.13).
- Claude Code for the agents, and tmux 3.2+ for the fleet.
- Node 22+, optional, for the codebase snapshot each session starts from.
The package is aisquare-cli; the command is aisquare, with asq as the
short alias.
Documentation
| Page | What is in it |
|---|---|
| Install | The one-liner's flags and exit codes, reading it before you run it, installing by hand, Windows, starting the UI, updating and removing it |
| Memory | Your agent remembers preferences and project conventions, and starts every session oriented. For everyone; nothing to run after setup |
| The fleet | The UI, the roles, aisquare fleet …, Claude accounts, and every default you can change |
| Orchestration | Several agent sessions work one problem as a team, with a shared task board. Opt-in, per repo |
| Reference | The hooks, the ~/.aisquare layout, the main command tree and the global flags |
| The AISquare platform | Optional: signing in, Explainability, the Collective Intelligence test bed |
Community
Questions and ideas go to Discussions, bugs to issues. Pull requests are welcome: CONTRIBUTING has the setup and the one command CI runs. Releases are in the changelog; to report a vulnerability, see SECURITY.
License
Metadata
Release files for aisquare-cli 0.8.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| aisquare_cli-0.8.0.tar.gz | 4.6 MB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| aisquare_cli-0.8.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 5.7 MB
Release files / aisquare_cli-0.8.0.tar.gz
| Download URL | aisquare_cli-0.8.0.tar.gz |
|---|---|
| Size | 4.6 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3cac1c8a8933bab6452b8bd0667c7f32ec2d2d21dde3d1abfff52a011499101a
|
|
BLAKE2b-256 checksum How to use checksums |
4780d67e69ab4558a6d3f77875ddcb4b9c024a013b83e593265551bd211548b1
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 13, 2026.
Transparency logRelease files / aisquare_cli-0.8.0-py3-none-any.whl
| Download URL | aisquare_cli-0.8.0-py3-none-any.whl |
|---|---|
| Size | 1.1 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
14d20c52d4d1aec30b9bbe9431de02d9701ab6119979536db463c88d7e8d9b57
|
|
BLAKE2b-256 checksum How to use checksums |
2b165540fb9c7d291b0aacfd4cc0f7fcc9f8b94f73cb3ec82730223d693379ec
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 13, 2026.
Transparency log