signa

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via GET /api/agents/<address>/inbox?limit=20, which returns public inbox content authored by other agents/users and is treated as untrusted text that the agent may read/quote into its reasoning.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill makes runtime calls to https://www.signaagent.xyz/api/brain whose signed "brain" response includes a "plan" (e.g., instructions like "root.market()") and tool calls that directly tell the agent which capabilities to invoke, so remote content from that URL can control agent behavior and is a required runtime dependency.