intercom

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly instructs calling Apideck's proxy at https://unify.apideck.com/proxy (via the x-apideck-downstream-url header) which returns Intercom's raw API responses — these responses can include untrusted, user-generated content (messages/conversations) that the agent is expected to read and could materially influence its actions.