agents-md

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

• Third-party content exposure detected (medium risk: 0.30). AGENTS.md/SKILL.md are loaded from the (audited) repository itself before user input, and in untrusted audit mode the workflow reads that attacker-controlled repository text (e.g., static checks over files and references) which can include outsider-authored free-form content.