check-upstream-updates

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly scrapes public upstream release notes and changelogs (Step 4: "GitHub Releases", "CHANGELOG.md", gh api repos/.../compare) and uses those untrusted, user-authored contents to decide triage and create GitHub issues, so third-party content can materially influence actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill issues runtime calls to GitHub (e.g., gh release list -R <owner>/<repo> and gh api repos/<owner>/<repo>/compare/v<installed>...v<latest> / gh api repos/<owner>/<repo>/security-advisories) and scrapes release bodies/CHANGELOG.md at runtime to drive its analysis, so remote content from those URLs is injected into the agent's decision/context and could therefore influence its outputs.