pull-request

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly calls out reading remote GitHub data (e.g., gh pr view in Step 1) and instructs the agent to "follow repository conventions if visible from recent PRs" (Step 2 of SKILL.md), which requires fetching and interpreting user-generated PR content from GitHub that could influence how the agent drafts or modifies the pull request.