Aluy beta

A single terminal that puts
dozens of agents
to work for you

It reads and edits your files, runs your commands and takes the task to the end — with the provider you already use, local models included. Your session, your data and your key stay on your machine.

Read the docs
install — one line
bash — install aluy
$ curl -fsSL https://aluy.dev/install.sh | bash
1 line · no sudo·installs to ~/.aluy·re-run to update

Aluy renders best in WezTerm: the icons and borders come out visually more refined. It works the same in other terminals, just plainer.

see it run

A real cycle: from broken test to green.

your first minute
$ aluy # opens a session in this project
/doctor ✓ model · key · MCP — all good
/init creates ALUY.md for this repo
/cycle "run the tests and fix what breaks"
working in cycles… stops on green or at the cap ↩
/memory list what aluy remembers here
one agent, at scale

What you'll put it to work on.

Tasks

Bulk renames, file wrangling, reports, data cleanup — describe it, it does it in your shell.

Processes · robotic automation

Wires steps and tools into one process. End to end, instead of brittle scripts that break.

Testing

Writes, runs and fixes until it passes. Reads the error, corrects it, runs again.

Development

Reads the project, proposes the change, edits files and runs the build. You review and approve.

what it does differently

Not a chat about your code. A team that acts.

/cycle

Kick it off and go do something else

Works in cycles on its own until the task is done — with a handbrake. Hard caps on time and steps, plus automatic stops.

/subagent

A fleet of specialists

Each subagent is a profile (.md) with a focus — one for tests, one for review, one for docs. Delegate and they work in parallel.

/rooms

Watch the fleet collaborate live

Rooms put several agents talking to each other about the problem — debating, splitting the work. You follow in real time and jump in whenever.

/mcp

Plug in what you already use

Through MCP — the open standard for connecting tools — the agent talks to databases, APIs and services. It can actually use them, not just talk about them.

the model's path

Local where it matters, connected only where it has to be.

stays on your machine

The session, history, project files, memory and your provider key. The agent runs locally, behind the permission gate.

crosses the edge

The model call goes straight to the provider you configured, with your key. No middleman and no metering.

turbocharge your terminal

Three optional add-ons, running on your own machine.

mem0

Memory that outlives the session: the agent keeps facts and preferences and recalls them in the next conversation.

headroom

Condenses the conversation as it grows, so a long session keeps fitting in the model’s window.

ollama

Runs small models on your machine for Aluy’s own internal work — those calls never leave it.

All three are optional and come with turbo mode. Without them Aluy works the same: memory stays in the project’s ALUY.md, written by you or by /init.

no overselling
Not an IDE. It runs in your terminal, on the project you already have.
Not a model. You bring your provider, local models included.
Still beta. Things break, and the changelog says what changed.