What we collect, and why

If you only visit the site

Our web server keeps an ordinary access log. It’s the same thing every web server keeps, and it’s the only record we have of a visit.

• WhatYour IP address, the time, the page you asked for, whether it worked, the browser identification string your browser sends, and the page you came from if there was one.
• WhyTo keep the service secure, to work out what broke when something breaks, and to see roughly how much traffic the site gets. There is no analytics service, so this log is the only way we can answer that last one.
• How long30 days, then it is deleted automatically. That is also the limit of how far back we can look.

We don’t set any cookie for a visitor who isn’t logged in, we don’t load third-party scripts, trackers or ad pixels, and nobody outside receives this log.

If you create an account

• WhatYour email address, a name to show you, and your password — which is stored only as a cryptographic hash, never in a form anyone here can read. If you answer the two optional questions when signing up (what you’re building, and what you most want to know about your code), we keep those answers too.
• WhyTo give you an account, let you log back in, and send you the emails an account needs — confirming your address, resetting your password. The optional answers help us understand who the product is for; nothing depends on them and you can leave them blank.
• How longFor as long as you have an account. Ask us to delete it and we will.

Once you log in we set one cookie so you stay logged in. It exists to keep your session and is not used to follow you anywhere.

When you accept the agreement

We record which version of the agreement you accepted, the exact text of that version, when you accepted it, your IP address and your browser identification string. This is the one place we keep an IP address deliberately and for a long time: a record of an agreement has to be able to show who agreed to what, and when, or it isn’t a record. It is kept for as long as the agreement record itself.

When you submit code

• WhatThe code you submit, and details about the submission — the file name and size, or the repository address you gave us.
• WhyTo analyze it. It is analyzed inside an isolated, single-use environment that is destroyed afterwards.
• How longThe submitted code is automatically and permanently deleted 30 days after you attach it. That deletion is enforced by the system on a timer, not by someone remembering.

Your report is kept so you can come back to it, and the report quotes parts of your code — the lines a check flagged, and what the check said about them, with any credential values masked out. Those quotations stay with the report after the submitted code itself is gone; that is what makes a finding auditable months later. The full details are in the agreement.

If you link a GitHub repository

• WhatA GitHub access token, issued to us by GitHub when you connect your account, and the GitHub username it belongs to. It can read the contents of only the repositories you chose to grant, and nothing else about your GitHub account. We never see your GitHub password.
• WhyTo fetch a snapshot of the repository you link, so it can be analyzed exactly as an uploaded file would be. The snapshot is then covered by everything in the section above, including the automatic deletion.
• How longThe token is stored encrypted and expires within eight hours. We delete it the moment you choose Disconnect, and we also delete it once it has expired, so an unused connection does not linger. You can revoke it at any time from your GitHub account settings under Authorized GitHub Apps, which works even if our service is unreachable.

Linking is optional. If you would rather not connect GitHub, upload the code as a file instead.

When you pay

Card details are entered on the payment provider’s own page. We never see, handle or store your card number. What we keep is the amount, the currency, and the reference numbers needed to match a payment to your order.

Who else sees your data

The Analyzer runs on infrastructure we control, so the default answer is “nobody outside”. Below is the complete, current list of outside companies that receive any of your data, what each one gets, and — just as important — what they never see. The short version: we don’t send your code outside Kind Robots unless you buy a service that requires it — and when a service does, this page says exactly what gets sent, we tell you before you buy, and we ask you to confirm.

Adding a company here is a disclosure event, not a quiet setting change — and so is widening what an existing one receives. This list only grows when it has to, and this page is updated the same day it does. The most recent change to what a vendor receives was the AI architecture review, on August 16, 2026.

• Anthropic

  Makes the AI model that writes your report's plain-language explanations, and the one that produces the AI architecture review if you buy it

  What they receive

  Two different things, for two different parts of the product. (1) Report wording, for every report we deliver — even for customers who never bought an AI service: a short, closed set of details about each finding, being its identifier, how serious it is, the industry-standard defect number, and a general issue category. (2) The AI architecture review, only if you buy it: a structural digest built from your project — file and folder paths, the names and declared versions of the packages you depend on, your database tables and columns, your routes, the names of the environment variables your code refers to, and short, length-limited excerpts of your code with credential values masked out.

  What they never receive

  Your source code in full — never a repository, an archive, or a complete file. The values behind your secrets: environment variables are sent as names only, and credential values are masked out of every excerpt before it leaves us. The original wording our checks produced. The report-wording part is narrower still: it receives only the finding details above, and nothing that identifies your project — file names, paths and package names are filled back in on our own systems after the AI writes its explanation, so that part never sends them at all.

  On the list since

  August 12, 2026 — architecture review added August 16, 2026

• Resend

  Delivers the account emails we send you

  What they receive

  Your email address, and the account emails sent to it: your signup confirmation, password reset emails, and email-change confirmations.

  What they never receive

  Anything about your code or your analysis — no source code, no findings, no report content, no report text. This vendor only ever handles account emails and has no connection to the analysis side of the product.

  On the list since

  August 15, 2026

• Stripe

  Takes your payment

  What they receive

  Your card details and the email address you enter on Stripe's own secure checkout page rather than ours — so we never see or store your card number. From us, Stripe also gets which service tiers you bought and internal reference codes, so we can match your payment to your order.

  What they never receive

  Anything about your code, your findings, or your report. Your account with us — as of August 17, 2026 we no longer send Stripe your account identifier at all. The tiers you bought reach Stripe only as prices we set up in advance, never as a description of what was analyzed or what we found.

  On the list since

  August 16, 2026

What each of these companies keeps, and for how long, is governed by that company’s own published terms — not by anything we can promise on their behalf. We say what we send them and what we never send them, because that part is ours to control.

You won’t find our database, our workflow engine, our security dashboard or our code-scanning platform on that list. That’s deliberate: every one of them is self-hosted, on our own infrastructure — we don’t send them anything, because there’s no outside company on the other end to send it to. Using a vendor’s software ourselves is not the same as sending that vendor your data, and this list is only credible if we’re honest about that distinction. Only the three companies above ever receive anything about you or your code.

We don’t sell your data, we don’t share it for advertising, and there is no advertising on this site to share it with.

Getting a copy, or asking us to delete it

Write to privacy@kindrobots.ai and ask. You can ask for a copy of what we hold about you, ask us to correct it, or ask us to delete your account and its data. A real person reads that address and will tell you what we did.

Two honest limits. Some things we can’t delete on request: the record of an agreement you accepted, and records of payments, both of which exist precisely so they can’t be quietly changed afterwards. And the access log above ages out on its own 30-day clock rather than being searchable by person — there is no index from you to those rows.

If this changes

This page can be updated, and it is meant to be — it describes what we actually do, so when that changes this changes the same day. The agreement works the other way: each version is fixed and identified by a fingerprint of its text, and changing it means publishing a new version rather than editing the old one. If the two ever seem to disagree, the agreement is what binds.