Privacy policy

Who the controller is

The data controller is SISIP SPA, a company incorporated and domiciled in Chile, which operates RepoFit. For any question about this policy, to exercise your rights, or to request a data processing agreement, use the email address at the end of this page.

• SISIP SPA · Chile
• Tax ID (RUT): 77.146.172-7
• Registered address: Lota 2258, Providencia, Región Metropolitana, Chile
• support@repofit.ai

What we store

When you connect a repository, RepoFit ingests metadata from its git history: commit identifiers, dates, the first line of each message, change counters, file paths, and the names and emails of the authors (needed to reconcile identities and build the team map). We also store what you and your team create on the platform (boards, specs, domain context) and the structured results your own AI agent sends over MCP (verdicts, modules, findings). From your account we store your email, your name, your profile picture if you upload one, and the last day you used the service (the date only: no time and no visit history).

What we do NOT store

• Your source code, ever. We do not store files, diffs, or fragments of your code. The analysis uses a temporary read-only copy that is deleted when it finishes, and the deep semantic analysis is run by your own agent in your environment. Two declared exceptions: we store the CODEOWNERS file (governance metadata: owners by area) and the structured results your agent chooses to send us — agents are instructed to reference paths, not to paste code.
• The body of commit messages (only the first line).
• Secrets, .env files, or credentials from your repos (they are never read).
• Access tokens from your social login (used only to verify your email, then discarded).

How we measure the product

To know whether RepoFit actually helps —and which recommendation improves a repository and which does not— we compute a numeric summary of what happened on the platform once a day. The summary carries neither your identity nor your workspace's: they are counts, not records. It is computed from information we already hold in order to provide the service —we install nothing new and we do not track you across the web— and that information is deleted along with your workspace. This is everything we count:

• How many workspaces had activity each day (a number, not which ones).
• How many times each type of action happened on the platform.
• How many times each rubric check was closed, and how many times it failed again.
• How many cycles were approved and how long they took.
• How many analyses finished and how many failed.
• How many repositories improved their score, and by how many points in total.
• How many times each MCP connector tool was called each day, and how many of those calls failed. This count is recorded at the moment of each call, without storing who made it, from which workspace, or what it asked for.

Nothing written by you or your team goes in: no card titles, no repository, branch or file names, no commit messages. Nor your IP, your browser, or tracking cookies. There is an automated test in our code that prevents adding any measurement not on the list above.

Purposes and legal basis

How long we keep each thing

Who else processes your data

To provide the service we rely on infrastructure and service providers acting as processors on our behalf. The list is published and kept current, with what each one does, what data it touches, and where it hosts it.

Aggregate statistics that survive deletion

When we analyze a repository we also store an anonymous measurement with no link back to it: the rubric version, the score, the band, how each pillar and each check came out, and the size of the code in bands. It carries no repository name, no workspace, no commit identifier, and not a single file path, and it cannot be re-associated with anyone — it isn't that we promise not to: the data that would allow it isn't stored. We use it to calibrate the rubric, which is the instrument that then measures everybody, and that is why those measurements are not deleted when you delete a repository or your account: by then they have stopped being your data and are statistics. Everything else is deleted, including the only part that reveals anything about your code: the findings with their paths and the quotes from your documentation.

People who appear in the history and are not our users

When you connect a repository we ingest the name and email of whoever made each commit, and from that we build the team map: we group one person's identities together and derive aggregate indicators of their activity in that repository (which areas they worked in, how often, what share of their changes are fixes). Those people are usually your team and have no account with us, so three points are worth being explicit about. First: that data is already in the git history you control, and we process it on your instructions to provide the service — in that processing you are the controller and we are the processor, with the commitments in the data processing agreement. Second: telling your team you use this tool is your responsibility, because it is your relationship with them and we have no way to contact them. Third: we do not build individual performance profiles, we do not score people, and the score is always the repository's, never anyone's; the per-person indicators exist to locate knowledge in the code, not to evaluate anybody. Any of them can exercise their rights by writing to us, and if you delete the repository their derived data goes with it.

International transfers

Our infrastructure is hosted in data centres outside Chile, and some of our processors operate from other jurisdictions. That means your data is transferred internationally. In those cases we require each processor to make contractual confidentiality and security commitments equivalent to the ones we make to you, and we limit processing to the purposes declared in this policy.

Your rights

You can exercise these rights at any time by writing to the email address at the end of this page. We will reply within thirty calendar days at most. There is no charge for exercising them, and you don't need to justify your request.

• Access: know what data of yours we process and get a copy.
• Rectification: correct inaccurate or incomplete data.
• Erasure: ask us to delete your data (you can also do it yourself, from the application).
• Objection: object to processing based on our legitimate interest.
• Portability: receive your data in a structured, commonly used format.
• Restriction: ask us to suspend a processing activity while a dispute is resolved.
• Complaint: go to the data protection authority that applies to you.

Your data, your control

• You can delete a repository and all its derived data from its page, at any time.
• You can delete your account and its data from Settings → Account and security.
• We do not sell your data, we do not transfer it to third parties for commercial purposes, and we do not use it to train artificial intelligence models. It is processed on our behalf by the processors we need in order to operate — infrastructure, the git providers you connect and, only while you grant support access, the AI tool we review it with — and each one is published with exactly what it touches.

Cookies

We only use cookies that are necessary for the service to work. We use no advertising, profiling, or third-party analytics cookies, so there is nothing to consent to or refuse: if you block them, the service stops working.

Storage in your browser

Besides cookies we keep two preferences in your browser's local storage. They never reach our servers, they identify nobody, and you clear them by clearing the site's data. We declare them because the rules aren't about cookies but about storing information on your device, and this is that.

When you connect your own AI agent

The deep analysis is run by the agent you choose (Claude Code, Codex, Cursor, or another) on your machine or in your CI. We hand it the questions over MCP and receive its answers, but the agent runs under your control and with your credentials: whatever you have it process travels to the model provider you contracted, under that provider's terms and privacy policy, not ours. We don't choose that provider, we have no contractual relationship with it on your behalf, and we cannot answer for what it does with that data. We say so because it is the part of the flow we don't control, and leaving it out would make this policy look like it covers more than it does.

Minors

The service is aimed at people who work professionally with code repositories and is not intended for minors. We do not knowingly collect data from minors; if we detect a minor's account, we delete it.

Security incidents

If a breach occurs that affects your personal data and may pose a risk to you, we will notify you without undue delay at your account email, describing what happened, what data was affected, and what we are doing about it, and we will report it to the authority where required.

Changes to this policy

If we change this policy, we update the date in the header. When a change materially affects how we process your data, we will notify you by email before it takes effect, with at least thirty days' notice, so you can decide whether to keep using the service.