Get started · about five minutes
From install to your first accepted change.
No account, no payment. Every command here comes from the agent's own repository, and the installer asks before it changes anything on your computer.
Setting up AlbertCode SWE Agent
Terminal, Browser, VS Code · command albertcode
Check the prerequisites
- Windows
- Windows 10 or 11, x64 (the x64 build also runs on ARM PCs). PowerShell 5.1 or 7.
- macOS
- A current macOS, on Apple silicon or Intel.
- Linux
- x86-64 with glibc: Ubuntu 20.04+, Debian 10+, Fedora, RHEL 8+, openSUSE, Arch. Not Alpine or other musl systems, and not ARM yet.
- Disk
- About 1 GB free.
- Network
- pypi.org and github.com while installing; your model provider while working.
- Git
- Recommended. The installer offers to install it if it is missing. Direct mode needs it.
- Python
- Not needed. The installer uses your Python 3.12 if present, otherwise downloads one for AlbertCode alone.
There is no official container image: the agents install directly for your user account. A model provider account (or a local model server) is the only other thing you need.
Install
The command downloads a small bootstrap script from the agent's repository. It finds the latest release, downloads the installer for your system, and checks it against the release's published SHA-256 checksums before running anything. Then the installer shows its plan and asks Proceed? [Y/N].
Install AlbertCode SWE Agent
for your user · asks before changing anythingcurl -fsSL https://raw.githubusercontent.com/- Open Terminal (⌘ Space, then type Terminal), paste the command, read the plan, then type Y.
- Open a new Terminal window when it finishes, then run albertcode --version.
Read before running, or pass options
Download installer/install.sh or installer/install.ps1, read it, then run it. Options: --check only shows the plan, --yes installs without asking, --version X.Y.Z installs a specific release.
curl -fsSL https://raw.githubusercontent.com/$env:ALBERTCODE_INSTALL_ARGS = '--check'; irm https://raw.githubusercontent.com/Verify the installation
Open a new terminal first. The installer added the command to your PATH, and only new terminals see it.
albertcode --versionPrints the installed version.
albertcode doctorWhich build tools AlbertCode can find, and where.
albertcode: command not found? Open a new terminal. Every installer run writes a log to ~/.local/share/albertcode-installer/install.log (macOS, Linux) or %LOCALAPPDATA%\AlbertCode\install.log (Windows).
Connect a model
The first time you start AlbertCode with no model connected, the connect step opens by itself. Pick a provider, paste its key — or give the address of a model server on your machine — and choose a model. The key is checked before it is saved, and kept by your system's keychain.
Hosted
An API key from a hosted provider, OpenRouter (many vendors through one key), Hugging Face, or any compatible service.
On your machine
Ollama (http://localhost:11434/v1), LM Studio (http://localhost:1234/v1), vLLM (http://localhost:8000/v1) or another local model server. With a local model, your code never leaves your computer.
albertcode models testSends one tool-calling request. The model must support tool calling to make changes.
Skipped it? Type /connect at the agent's prompt any time, and /model to switch.
Run your first task
# open a project, then start the agentcd your-projectalbertcode
Then describe what you want, in plain words:
- 1.It reads the repository and proposes a plan. Nothing is written yet.
- 2.You approve the plan — or ask for changes, or reject it.
- 3.It builds the change in an isolated copy, runs your tests and reviews the result.
- 4.You read the diff and accept it. Only then does the change reach your files.
Prefer the browser? Run albertcode web: it opens the same session, signed in.