Legal
Privacy Policy
Last updated: August 30, 2026
PromptShield is an AI prompt security inspector that runs a neural guardrail over every submitted prompt before allowing it to reach a model. This policy describes what data we collect and how it is used.
What we collect
- Your prompt text. The text you submit is sent to an AI model via Google Gemini to verify the guardrail pipeline. It is processed by the selected model provider under their terms and is not used by us for anything else.
- Your email address, only if you choose to log in. We use it solely to send you a magic login link and to identify your account. Auth is handled by Supabase.
- Saved stacks. If you are logged in and click save inside the Forge workspace, the idea and generated plan are stored in our database (Supabase), linked to your account. Only you can read or delete them; access is enforced with row-level security.
What we do not do
- No selling or sharing of your data with third parties.
- No advertising, no tracking cookies, no analytics profiles.
- No storage of anonymous guardrail runs.
Service providers
We rely on Supabase (auth and database), Google Gemini (model dispatch after the guardrail passes), and Vercel (hosting). Each processes data as needed to provide their service, under their own privacy policies.
Deleting your data
You can delete any saved stack from the Forge history at any time. To delete your account and all associated data, contact us via the GitHub repository.