market-sizing

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s workflow explicitly requires invoking data-analytics:user-context and then performing “live source reads” across public sources (e.g., public web content) to gather inputs; those runtime-fetched outsider-authored texts can be ingested into the agent’s LLM context during evidence gathering.