hermes-profile-audit

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

• Third-party content exposure detected (high risk: 0.70). The workflow explicitly audits “available sessions, logs, outputs, corrections, and handoffs,” which are outsider-authored behavioral evidence (e.g., messages/logs from other users or systems) and would be read as free-form text into the agent’s LLM context at runtime.