pm-architect

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

• Third-party content exposure detected (high risk: 0.90). The skill clearly ingests user-generated third-party content: scripts/delegate_response.py (and scripts/generate_daily_status.py / generate_roadmap_review.py) call the gh CLI to fetch GitHub issue/PR bodies and comments and then feed that content into autonomous Claude/auto-mode workflows, meaning untrusted web content is read and can directly influence agent decisions and actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). scripts/delegate_response.py calls the gh CLI at runtime to fetch GitHub issue/PR content (via GitHub API endpoints / URLs such as https://api.github.com or the repo issue URL https://github.com/rysweet/MicrosoftHackathon2025-AgenticCoding/issues/1504) and then injects that fetched content directly into a delegation prompt which is executed by the amplihack/Claude auto-mode, so remote content can directly control agent instructions.