clips-studio

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow ingests the outsider’s free-text prompt/motion instructions directly into scripts/falvid.py as the prompt argument for the fal.ai generation call (text-to-video cmd_generate and video cmd_animate/cmd_camera).

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged several high-entropy, literal tokens embedded in the test file. They match the fal key pattern (UUID:id-secret form) used by the code's FAL_KEY_RE and look like real API-style credentials (e.g. SAMPLE_KEY, ADMIN_KEY and other id:secret strings). Even though they appear in unit tests as sample/test values, they are high-entropy literals that could be mistaken for real credentials or accidentally committed; treat them as secrets and redact/remove them (or replace with obvious placeholders) in the repo.

No potential matches were ignored — the repo contains these explicit literal tokens which meet the "high-entropy, literal value" rule.