new

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

• Third-party content exposure detected (high risk: 0.80). The skill's get-notes.sh (and SKILL.md Step 6) explicitly fetches PR/MR titles via gh/glab/tea and reads commit messages from the repository (user-generated, potentially untrusted content) and instructs Claude to read/post-process those notes and use them in release notes and changelog, so third-party content is ingested and can influence actions.