using-agent-relay

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to include workspace keys and agent tokens directly in example tool calls, CLI flags, and environment exports (e.g., "rk_live_..." and "at_live_..."), which requires the LLM to handle or echo secret values verbatim in outputs — a high exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow has the participant call check_inbox(limit: 20) and read channel/message history via tools like check_inbox, list_messages, get_message_thread, and search_messages, which ingests outsider-authored free text posted by other users into the Relay workspace/channels.