using-agent-relay

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt exposes a workspace_set_key tool ("Set the workspace API key") that implies the agent will receive and must pass an API key value verbatim as an argument, which creates an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime “Receiving Messages” path (e.g., mcp__relaycast__message_get / mcp__relaycast__thread_get / mcp__relaycast__message_search) ingests message body text authored by other agents/users (e.g., “Relay message from Alice …: Content here”) into the agent’s LLM context, enabling indirect prompt injection via outsider-authored free text.