groundgate decides whether extracted facts can be trusted, so a way to make it decide wrongly is a security bug. Examples:

• a document, schema or candidate that makes admit admit a value the spec rejects, or decide differently on the same inputs;
• a receipt that verify accepts although its inputs or decisions changed;
• document or candidate text that runs script or escapes its element in the report page.

A model extraction that passes every check and is still wrong is not a vulnerability. It is the documented limit in the README's "What it does not do".

Report a vulnerability privately through GitHub: the repository's Security tab, Report a vulnerability. Please don't open a public issue. I aim to reply within a week.

Supported: the latest 0.x release.