feishu-seedance-video-pipeline

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow explicitly downloads and uses external HTTPS reference URLs and polls the Seedance (Volcengine) API as part of normal operation (see "Publish references" and "Seedance API Essentials" plus Reviewer Module gates that require reading payloads, reference URLs and API GET responses), which means it ingests untrusted/public third‑party content (arbitrary URLs and service responses) that can materially affect prompt construction, review decisions, and submission actions.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The prompt explicitly instructs running sudo commands (mkdir/cp/chmod into /var/www) and suggests disabling browser sandboxing, which directs privileged filesystem changes and security-bypassing actions on the host.