pull-request

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow reads GitHub review thread/comment bodies via ./scripts/gh-review-comments (GraphQL reviewThreads -> comments.body) before generating resolve/reply/comment actions, so outsider-authored PR review text can be ingested.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs the agent to "escalate network permissions for gh when sandboxing blocks GitHub API calls," which directs bypassing sandbox/network restrictions and thus encourages compromising the runtime environment.