Your data in Kameleo.
This notice describes the prototype’s data flow. This version describes the proposed S3 architecture.
Accounts and practice data
When authentication is enabled, Clerk handles sign-in. Kameleo associates your Clerk user ID with your attempts, edited challenge files submitted for grading, chat messages, and test results. Draft edits are saved in browser localStorage; clearing site data or changing browsers can lose an unsubmitted draft. Discussion posts and replies are stored with your account and are visible to other signed-in users under a generated member label.
AI and code execution
Using the assistant sends your prompt, relevant conversation history, and challenge code context to Google’s Gemini API. Submitting a challenge stores changed files in MongoDB and queues its ID through RabbitMQ. A separate Celery worker runs the Docker grader without network access. Run Tests sends your current edits through the same queue for a practice run and stores the latest test results and console output on your attempt without submitting it. Read-only challenge templates come from S3. Submitting an open-ended implementation also sends its prompt, source files, and design notes to the configured AI provider to generate tests and an advisory review. Generated checks, review text, and recorded execution results are stored with the attempt.
Do not paste passwords, API keys, confidential work code, or personal information into challenges or chat. Provider processing and retention also depend on their policies and the account settings used by Kameleo.
Cookies and operational logs
Authentication uses session technology, including cookies. Hosting providers process requests to deliver the service. Server access logs record requests and response status. Authentication tokens are sent in WebSocket frames rather than URL query strings.
Retention and requests
The prototype does not yet have an automatic deletion schedule or a self-service data export or deletion tool. Before submitting personal data, ask the person who invited you to the prototype about retention or to arrange access, export or deletion. A public privacy contact must be established before open registration.
Provider information
Learn more from Clerk, Google Gemini, AWS, and MongoDB. Which services are active depends on the prototype’s deployment configuration.