dart

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow uses ClawLink to “connect dart” and then the agent runs Dart actions via the ClawLink CLI; this necessarily involves fetching/reading Dart-provided content (e.g., action metadata/results) that is authored by an external third-party service, which can be free text ingested into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires running "npx @useclawlink/cli" (package at https://www.npmjs.com/package/@useclawlink/cli), which will fetch and execute remote code at runtime to control actions/flows, making it a required external runtime dependency.