dragonruby

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md includes an "Async HTTP" runtime example showing args.gtk.http_get('https://example.com/data.json') and subsequent JSON parsing, which demonstrates fetching and ingesting arbitrary public URLs (untrusted third-party content) that the agent is expected to read and could materially influence game behavior.