reviewdog

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

• Third-party content exposure detected (medium risk: 0.65). At runtime, the required workflow pipes the output of security tools (e.g., semgrep ... --json, bandit ... -f json, gitleaks ... --report-format json) into reviewdog, and those tool outputs include free-text fields like message/diagnostics derived from the scanned repository contents (outsider-authored PR code), which reviewdog then embeds into LLM context via its comment-generation path.

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

• Potentially malicious external URL detected (high risk: 1.00). The GitLab CI template includes a runtime "curl -sfL https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh" which fetches and executes a remote install script, and the GitHub Actions workflow uses the external action "reviewdog/action-setup" (github.com/reviewdog/action-setup) which is fetched and executed at runtime — both are runtime-executed external code dependencies required by the skill.