policy-starter

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

• Third-party content exposure detected (high risk: 0.85). The skill’s REQUIRED workflow includes “Web search for the current published model policies and guidance” and then “read the model policies” from public sources (ABA/state bars/ILTA/CLOC/peer-firm policies), which are outsider-authored free text fetched at runtime and ingested into the agent’s LLM context for drafting.