agent-skills-evaluator

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

• Third-party content exposure detected (high risk: 0.80). The skill documentation explicitly states the Agtm CLI "can download skills from GitHub" and that the prepare step "Fetch the evaluator and benchmarks.json", meaning public, user-generated content from GitHub/public files is ingested and can influence agent behavior and subsequent actions.