ljg-qa

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

• Third-party content exposure detected (low risk: 0.10). The workflow reads external articles/papers via WebFetch or PDF reading based on user-provided URLs or files, which are active user-selected inputs rather than monitored feeds.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill contains an operational curl command that fetches and executes code/instructions from a local endpoint during runtime, presenting a weak/suspicious external dependency risk.