seedance-storyboard
Pass
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: SAFEDATA_EXFILTRATION
Full Analysis
- [Data Exposure & Exfiltration] (LOW): The script communicates with an external API at
api.xskill.ai, which is not on the trusted whitelist. It sends user-provided data (prompts) and can read local files via the--image-pathargument to upload them, which is a common pattern for such tools but technically a data exposure surface. - [Indirect Prompt Injection] (LOW): The skill is susceptible to indirect prompt injection because it processes untrusted data from both user input and remote API responses.
- Ingestion points: Command-line arguments (
--params,--image-url) and JSON responses from the remote APIapi.xskill.ai. - Boundary markers: None present in the prompt interpolation or output display.
- Capability inventory: Network requests (GET/POST) via the
requestslibrary and local file read access for image uploads. - Sanitization: Validates JSON structure using
json.loads()for inputs, but does not sanitize the content of the strings processed or returned.
Audit Metadata