diagnose

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

• Third-party content exposure detected (medium risk: 0.30). The skill defines a debugging workflow (“diagnose”) that may use user-provided artifacts and prompts (e.g., error messages, commands/output, captured logs/HAR files) at runtime, so the outsider-authored free text enters the agent’s reasoning loop via the user’s bug report and supplied reproduction details.