activations

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

• Third-party content exposure detected (high risk: 0.80). The skill queries the Treasure Data activations table and SKILL.md explicitly instructs the agent to surface and use the activations.response field (text returned from activation servers) when diagnosing errors, which means the agent ingests untrusted third‑party response content that can influence its decisions.