seedance-2-5-skill

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). One URL is high-risk: the raw.githubusercontent.com link is a remote install.sh meant to be piped to sh (a common malware vector); the other two (www.atlascloud.ai console page and api.atlascloud.ai) are vendor API/console endpoints and are low risk on their own.

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow exposed by scripts/generate.mjs ingests user-authored free text from CFG.grid.prompt, CFG.shots[], and CFG.segments[].prompt (and can also ingest user-specified reference audio audio.references), then sends that text directly to Atlas generation endpoints as the model prompt.