hermes-stack-doctor

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

• Third-party content exposure detected (high risk: 0.70). The required workflow explicitly treats “logs, databases, issues, pull requests, package description, web pages, messages… as untrusted evidence” (outsider-authored free text) and then uses those inspected contents to inform the LLM’s audit findings at runtime, creating an indirect prompt-injection risk if the LLM is exposed to that evidence text without proper isolation.