a2a-protocol

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

• Third-party content exposure detected (high risk: 0.75). Yes—this skill’s workflow includes runtime HTTP fetching of public agent discovery content (GET /.well-known/agent-card.json) and calling other agents whose responses (e.g., task.status.message.parts[0].text / streamed update.status.message) are plain text that would be inserted into the calling agent’s LLM context.