adversarial-authoring

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

• Third-party content exposure detected (high risk: 0.85). The required workflow explicitly gathers “User request” and “Relevant OpenSpec instruction JSON” as task context and then sends that context to the adversarial-reviewer subagent, so the LLM context is fed with user-authored free text (outsider-authored relative to the operating user’s chosen inputs) via the “User request” path.