Acceptable use

These rules cover the website, the desktop apps, and hosted jobs. They sit beside the terms. If a job would break them, do not start it.

Work you may change

Use Crawn on projects you own or are allowed to change. That includes the repository, the branch, and any service the job can reach with credentials you supply. Do not point a job at a machine, account, or codebase you are not authorized to modify.

No attacks

Do not use Crawn to break into systems, probe them without permission, steal credentials, bypass access controls, or write or hide malware. Asking the model to draft an exploit against a target you do not own is the same rule.

Other people’s data

Do not place someone else’s private data, secrets, or production credentials into a prompt or a job unless you have the right to use that data for that purpose. A cloud model can receive the prompt and the files that job needs. Strip keys and personal data you do not need before you send the job.

What you still review

Crawn can be wrong, and it can follow a harmful instruction if you give it one. You review the diff and the checks before you treat the work as done. You are responsible for what you merge, run, or ship.

The hosted service

Do not overload hosted jobs, accounts, or APIs in a way that blocks other customers. Do not resell access, share a password, or automate signups to get around plan limits.

If a job breaks the rules

We can stop a job, remove hosted output, or suspend the account. We will say why when we can do that without helping the abuse continue. Serious abuse can be reported where the law requires it.