FAQ
What is it? A coding agent on your machine, with a terminal interface and a browser workspace on one local service, carrying the AlbertCode engine. It is proprietary software, distributed through this repository's releases; see LICENSING.md. Model use is billed by your model provider, or costs nothing with a local model.
Does it send my code anywhere? Only to the model provider you connect, and to MCP servers you add. No telemetry, no account. See Privacy.
Can it change my files without asking? Only in Direct · auto-approve, which you choose. In Governed, your files change only when you apply a reviewed change; in Direct · ask each write, each write waits for you.
Do the terminal and the browser have to stay open? No. Tasks run in the local service; close either and they keep going. Open either again to see them.
Can I use it beside AlbertCode SWE Agent? Yes. mini-AlbertCode has its own command, its own port (8778), and its own folders, keys and sessions; nothing is shared between them.
Which model should I use? One that supports tool calling. OpenRouter is a good start: one key for
most of the market. albert-mini models test checks a model.
How do I update? albert-mini update, or run the install command again. Your sessions are kept.
How do I remove it completely? albert-mini uninstall --all.
Where is the source code? It is not published. This repository has the installers' bootstrap, the documentation and the releases.
From docs/faq.md in huynvic/mini-albertcode-swe-agent at 59127ad (13 Oct 2026), shown unchanged except for links. mini-AlbertCode SWE Agent documentation, CC BY 4.0. Latest on GitHub.