LEAP

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

• Third-party content exposure detected (low risk: 0.10). The workflow retrieves external skill files via online searches and raw downloads (skills.sh and GitHub raw) during exemplar discovery and skill fusion, ingesting untrusted third-party free text.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill's runtime setup and execution pipelines fetch remote files directly from third-party repositories via GitHub raw URLs and skills.sh sitemaps, which constitutes a weak, unauthenticated, and uncontrolled external dependency tier.