synthesis-code-integration

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to fetch and review external contributor branches/PRs (e.g., "Fetch the branch and review the diff", "git fetch origin", "git checkout contributor/branch", and GitHub cherry-pick workflows), which requires ingesting untrusted, user-generated code from forks/PRs that the agent must read and that can directly influence integration and merge actions.