changelog-generation

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

• Third-party content exposure detected (high risk: 0.85). The skill fetches commit subjects and full commit messages from public third-party sources at runtime (e.g., Chromium via https://chromium.googlesource.com/.../+log/... and GitHub via https://api.github.com/repos/.../compare/...), then those free-text commit messages are passed into the agent’s summarization LLM context—an indirect prompt-injection risk from outsider-authored content.