code-review

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow is code/PR review oriented (e.g., “Remote PR number or URL” → checkout/read PR description, existing comments, and changed files), which commonly ingests outsider-authored free text from PR descriptions/comments and diff contents into the agent’s LLM context.