llm-gateway

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

• Potentially malicious external URL detected (high risk: 0.70). The skill pulls and runs external container images at runtime (e.g., ghcr.io/berriai/litellm:main-latest, postgres:16-alpine, redis:7-alpine), which execute remote code and are required runtime dependencies, so they present a supply-chain execution risk.