html-skills-listen

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

• Third-party content exposure detected (high risk: 0.80). This skill runs a localhost HTTP receiver (server.js) that accepts POSTed JSON bodies from interactive HTML artifacts and forwards the raw submission into the agent via notifications (see server.js HTTP listener and SKILL.md step 5 / server-start flow), meaning untrusted/user-generated submission content is ingested and explicitly used to continue tasks.