gold-price

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly instructs the agent to web_fetch data from the public API http://49.235.61.172:3003 (including /api/summary and /api/ai/snapshot), and the agent parses and acts on that returned (potentially untrusted) content — including a natural-language snapshot — which could materially influence its outputs and behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill issues runtime web_fetch calls to http://49.235.61.172:3003 (notably /api/summary and /api/ai/snapshot), and the /api/ai/snapshot endpoint returns natural-language content that would be injected into or directly control the agent's responses, so this is a runtime external dependency that can control prompts.