twingate-idfw

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime path is the self-hosted Twingate Gateway/web-app integrations that receive and verify user-presented request headers/JWTs (e.g., Authorization: Bearer {{jwt}} or trusted headers) coming from outside clients, which are outsider-controlled free text and are ingested by the Gateway/middleware at request time.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to inspect the remote repository URL https://github.com/Twingate/gateway at runtime for exact YAML/Helm values, so fetching that repo would directly influence the agent's answers and thus is a runtime external dependency.