Privacy
Last updated
This page says, in plain words, what repo-sensei keeps about you and your code, why, who else handles it, and for how long. repo-sensei is in early access; changes to this page are listed in the changelog.
What is kept
- Your GitHub account: its id, login, name, email and avatar, from signing in, and the GitHub tokens that let repo-sensei act for you (encrypted).
- Your sessions: when each started and was last used, and the address and browser it came from. Settings → Sessions shows them and signs any out.
- What you make: organizations, memberships and invitations, questions and answers, diagrams, fixes and changes, settings, and integrations (their secrets encrypted).
- The code of repositories you connect: a copy of each indexed commit, and the index made from it (symbols, search indexes, the code's graph, summaries).
- Usage: what each action cost and on which model, to keep within limits and caps.
- How fast pages were: in the app (not the public site), your browser reports how long a page
took to show and to respond, with the page's route as a pattern (
/[org]/[repo]/ask). It holds no names, and nothing that says who you are beyond your organization. These are kept 30 days and shown, combined, on your organization's insights. - The audit log: sign-ins and changes to members, roles, repositories, keys, integrations and fixes, with who made them, when, and the address they came from.
For questions asked without an account on Explore, no account is made. The address they came from is kept only as a keyed hash, to apply per-address limits.
Who handles it
- AI providers, to index and answer: parts of your code and your questions go to Anthropic (answers, summaries, plans and patches) and to Voyage AI (embeddings and reranking).
- GitHub, which repo-sensei calls to read the repositories you connect and, for a fix or change you approve, to open a pull request.
- Email: invitations and the notifications you choose are sent through Resend, which receives the recipient's address and the message.
- Hosting: the service runs on one server at Oracle Cloud: the web pages, the API, its database and queues, and the isolated sandbox where a fix runs your repository's tests (reaching nothing but the package registries its dependencies come from). Cloudflare carries its traffic (HTTPS, DNS) and stores its database backups. Each stores or carries data only to run the service.
- Sentry and Slack, only when your organization connects them, to start fixes from errors and post notifications.
- Error reporting: errors in the service are reported to Sentry without personal data.
Nothing is sold, and there is no advertising or cross-site tracking.
Cookies and storage
A session cookie keeps you signed in, and a second protects against cross-site requests. Your browser keeps your appearance and layout choices itself (local storage); they aren't sent to the server.
How long
- Indexes of older commits are pruned; what a branch or pull request points at is kept while it is tracked.
- A deleted organization can be restored for 7 days, then everything in it is purged.
- A deleted account loses its sessions, GitHub tokens and its name, login, email and avatar at once. What it wrote in organizations stays with them, by an anonymous author. The audit log is never edited: its entries keep the anonymous account's id and the addresses they came from.
- Backups of the database are kept for 7 days, so anything deleted is gone from them within 8 days.
Your choices
Disconnect a repository to remove it and its index; delete an organization or your account from its settings. To ask what is kept about you, to have something corrected or removed, or for anything else about your data, write to privacy@reposensei.com.