Skip to content

Security: kadireren7/patchfrog

Security

SECURITY.md

Security Policy

Supported versions

PatchFrog does not yet have a tagged release train -- main is the only supported version. Run patchfrog ops doctor (its deployed_commit line) or git rev-parse HEAD to identify exactly which commit a given deployment is running.

Reporting a vulnerability

Do not open a public GitHub issue for a security vulnerability or for anything that includes secrets.

GitHub's private vulnerability reporting feature is not yet enabled on this repository. Until it is:

  1. Open a GitHub issue with a minimal, non-sensitive title and body (e.g. "requesting a private security contact channel") -- do not include exploit details, proof-of-concept code, or any secret in that issue.
  2. A maintainer will follow up to establish a private channel for the actual report.

If private vulnerability reporting has since been enabled for this repository (check the repository's Security tab), prefer that instead of the issue-based fallback above.

What counts as in scope

  • The PatchFrog application itself (patchfrog/, apps/) -- webhook signature verification, GitHub App authentication, publication safety gates, secret handling, and the AI review pipeline's trust boundaries (see docs/deployment.md's "Provider/model selection" and docs/operations.md's "Known security-boundary limitations" for what's already documented as a known, judged-low-risk gap).
  • Deployment configuration guidance in docs/ that, if followed, would result in an insecure deployment.

What's out of scope

  • Findings that require an attacker to already control a repository PatchFrog is configured to review (see docs/operations.md's "Known security-boundary limitations" -- the installing party already controls their own repository content).
  • Vulnerabilities in third-party dependencies -- report those upstream; a PR bumping a vulnerable pinned dependency here is still welcome.
  • Findings against PatchFrog Cloud -- it does not exist yet (see docs/external-beta.md).

There aren't any published security advisories