soultrace

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

• Third-party content exposure detected (high risk: 0.75). The required workflow calls an external, public API at runtime and ingests the response’s free-text question field (e.g., question.text) into the agent/LLM context, which is outsider-authored content.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill issues runtime POST/GET calls to https://soultrace.app/api/agent which return question text and next-question logic that directly control the agent's prompts and flow.