review-proposal-auto

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

• Third-party content exposure detected (high risk: 0.85). The skill reads an outsider-authored proposal artifact from docs/threads/<thread>/proposals/<...>-proposal.md (user did not author/choose the proposal text), and that proposal body becomes LLM-readable prose during step 3 “Read the proposal READ-ONLY,” which is then used to draft the review report.