agent-relay-chatgpt-setup
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill configures the agent to subscribe to and process external messages via the
relay.messageevent, which introduces a risk of indirect prompt injection from untrusted external agents. - Ingestion points: Data enters the agent's context through the
relay.messageevent subscription and the output of therelay_events_statustool. - Boundary markers: There are no instructions to delimit external content or treat it as untrusted data.
- Capability inventory: The skill utilizes tools including
relay_events_status,answer_reply_question, andopen_shared_sessions. It also explicitly enables automated responses to external agents ("Replies to agents that message you go out without asking"). - Sanitization: No evidence of sanitization or validation of the external message content is provided in the instructions.
Audit Metadata