vertex-engine-inspector

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and validates external agent endpoints (e.g., testing the AgentCard at AGENT_URL/.well-known/agent-card and calling the A2A Task/Status APIs as shown in SKILL.md and scripts/inspect-agent.sh), which are untrusted third-party HTTP endpoints whose returned metadata is parsed and used to determine compliance and next actions.