How we handle your code

Repoclaw needs to understand your product, not to keep your code. These are the rules it follows.

Reading

• Access is read-only. Through GitHub, the Clawpy Repoclaw app asks only for read access to contents, issues and metadata, and only on the repositories you choose.
• Or use local mode: Repoclaw runs on your own machine and uploads only the capability list. Your code never leaves your computer.
• Repositories are treated as untrusted. Repoclaw never runs your code, your build scripts or your git hooks.

Processing

• Your code is read on our server in the EU, or on your machine in local mode.
• Source code is never sent to an AI model. The model sees a summary: folder and file names, your documentation and roadmap, dependency names, and TODO comments.
• Summaries go only to model providers whose terms exclude training on your data. Today that is Mistral AI, hosted in the EU, with training switched off.

Storage and deletion

• Clones are deleted at the end of every scan, including scans that fail.
• We keep your capability list, your weekly reports and the verdict history that makes week-to-week comparison possible.
• You can delete everything at any time from your dashboard or by emailing support@clawpy.ai. Backups roll off within 30 days.

What we never do

• Sell or share your code or reports.
• Use your code to train models.
• Publish a report about your repository without your permission.