ops-chutes

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

• Third-party content exposure detected (high risk: 0.80). The skill calls external services (INFERENCE_API_BASE "https://llm.chutes.ai" and MANAGEMENT_API_BASE "https://api.chutes.ai" from util.py) and manager.py directly consumes their responses (e.g., list_models/list_inference_models, timed_ping, get_chute_details, get_usage_history) to drive logic and recommendations (notably the recommend command), so untrusted third‑party API responses can materially influence agent decisions.