Data Use

• Your code — files are read and edited locally. We never upload your source code to our servers.
• Workspace settings — your project configuration, rules, and preferences live in a local .atlarix folder inside your project.
• Conversation history — chat sessions are stored locally in an SQLite database on your machine.
• API keys you bring — stored in your OS keychain (macOS Keychain, Windows Credential Manager, or the system secret store on Linux). Never sent to Atlarix servers.
• Local model inference — when you connect Ollama, LM Studio, or any local endpoint, all inference stays on your hardware.

• Atlarix Core — unless you set up your own models, Atlarix runs on Atlarix Core. Your request passes through our authenticated proxy, which verifies your account before anything is sent — we don't log or retain the request. From there it goes directly to the model provider that serves it, who processes it under their own data policy. We don't have zero-retention agreements with these providers, so we can't guarantee what they keep. See our Models page for how models are connected in Atlarix.
• Atlarix Auto (default) — Atlarix Auto picks the Atlarix Core model for each step, and it is where new installs start. With Atlarix Pro or credit, Auto is paid and handled exactly like any other Atlarix Core request — it is never shared for model training. In the desktop app only, an account with no credit can use Auto for free: free Auto runs on OpenAI models under OpenAI's data-sharing program and on free models reached through OpenRouter and Requesty, so the prompts, code and replies it handles are shared with the provider that serves each request, which may use them to improve its models. Some free Auto requests are instead served by open-weight models (such as MiniMax and Kimi) that Atlarix runs on Amazon Bedrock (Amazon Web Services); AWS processes those requests on our behalf, does not use them to train models, and does not share them with the model makers. Atlarix Reviewer's free requests go only to OpenAI. You are asked to accept this before any free use, and you can avoid it at any time with Pro, credit, your own API key or a local model. The Chrome extension never uses free Auto. Atlarix Reviewer uses free Auto only for public repositories, on an account with no credit.
• Local models (you set up) — run a local model — for example through Ollama — and your code and prompts never leave your machine. Nothing is sent, stored, or logged — by us or anyone.
• Your own API key (you set up) — add your own provider key and requests go directly to the provider you chose. Atlarix acts as a local client; we never see the traffic.

Training: Atlarix never trains on your code, in any mode. The one case where a model provider may use what you send to improve its models is free Atlarix Auto, described above — opt-in, in the desktop app (and Reviewer on public repositories) only.

Our hosted infrastructure stores only account and billing data — never code.

We do not

• Train on your code or prompts (a provider may only for free Atlarix Auto, which you opt into)
• Log or retain prompt content on our proxy
• Store your source files on our servers
• Share your data with third parties for advertising
• Sell your personal information
• Collect product analytics without your consent — it is opt-in, and off until you say yes

One thing that setting does not cover: crash and error reports are sent whether or not you opt in to product analytics. They carry stack traces and device information — never your code, prompts, or file contents — and we use them only to fix crashes. The app says so at first run and in Settings.