respond-to-pr-comments

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

• Third-party content exposure detected (high risk: 0.85). Yes—Step 2/3 fetches outsider-authored PR review thread comment bodies from GitHub/Azure DevOps APIs (e.g., reviewThreads.nodes.comments.nodes.body / comments[*].content) and then Step 5 uses that text in the agent’s analysis and draft replies, which becomes LLM context.