Security policy
Reporting a vulnerability
Please report security problems privately, not in a public issue.
Use GitHub's private vulnerability reporting: report a vulnerability, or choose Report a vulnerability on the repository's Security tab. Only the maintainers can see the report.
Please include:
- the affected file or component, and the version or commit;
- steps to reproduce, and what an attacker could achieve;
- any suggested fix.
What is in scope
- The bundled scripts in
skills/form-and-force/scripts/. They run on users' machines against their projects. The design rules: no network calls beyond the URL you pass toaudit_page.mjs, never read secret files, and never print a matched secret's value. - The repository tooling in
tooling/, the documentation site builder insite/, the brand renderer inbrand/, and the GitHub Actions workflows. The only workflow with write permissions is the manual Pages deployment, and the validator keeps it manual. - Skill instructions that could lead Claude into unsafe actions, such as installing packages globally, exposing secrets or fabricating claims.
Third-party websites cited in research/ are out of scope. Report problems with those sites to their owners.
Handling secrets
- Never include API keys, tokens, credentials,
.envfiles or private project files in an issue, a pull request, a log or a commit. - CI runs with read-only permissions and uses no secrets. The optional paid evaluations run only on your own machine, with your own credentials.
- If you commit a credential by accident, revoke and rotate it at the provider right away. Deleting the file afterwards is not enough, because it stays in git history and in any clone.
Limits of the automated checks
tooling/validate_repository.py and tooling/release_check.py scan for common credential patterns. A pattern scan cannot prove that a repository contains no secrets. It reduces risk; it does not replace review.
Supported versions
Security fixes are made on the latest release only.