SECURITY AND CONFIDENTIALITY
Security and confidentiality
Letting someone into your servers is a big decision. Here is exactly how we work, what our AI agents can and cannot do, and where your data goes.
Access: as little as possible
We ask for read-only access to your servers and code repository, with credentials created for us alone. You can revoke them at any time, and we hold separate credentials for each client.
- No direct write access to production
- Every change proposed as a pull request you or your developer can read
- Updates applied to staging first, with your tests
- A person approves every merge and every reboot
- A snapshot before every change, so it can be rolled back
- A full log of what was done, when, and by whom
What the AI agents see
Our agents read what they need for the task: package lists, dependency files, code, configuration and the relevant logs. Before any file is sent to an AI model, secrets such as passwords, API keys and private keys are detected and redacted.
The agents run in a sandbox and cannot act outside it. Content they read, including text in your repository or logs, is treated as data and never as instructions, and nothing they produce reaches production without human approval.
Which AI, and where
We use Claude, made by Anthropic. Under the commercial terms of the providers we use, your data is not used to train their models.
If your data must stay in Europe, we offer an EU option: Claude runs on Amazon Web Services (Bedrock) in an EU region, with requests kept inside that region. Data residency is not the same as zero retention, so the exact retention terms are written in your data processing agreement.
Contract and GDPR
We sign a data processing agreement (GDPR article 28) with every client. It lists our sub-processors, the region your data is processed in, how long it is kept and how it is deleted when you leave.
The service itself is a best-efforts obligation with a written scope: the systems covered, patch delays by severity, and exclusions such as zero days, software its maker no longer supports and targeted attacks.
When you leave
Every change we made lives in your own repository, so you keep all of it. We hand over the documentation, our access is removed and the credentials you gave us are deleted.
Sub-processors
The providers involved, depending on the option you choose. The data processing agreement holds the binding list.
| Provider | What for | Where |
|---|---|---|
| Anthropic | AI model (standard option) | United States |
| Amazon Web Services (Bedrock) | AI model (EU option) | EU region |
| Cloudflare | Website hosting, email routing, storing audit requests | Global network; audit requests stored in the EU |
Found a security issue on our site?
Please write to security@patchtempo.com. Our security.txt has the details.