tvc-transition-master
Pass
Audited by Gen Agent Trust Hub on Mar 25, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It reads content from a user-provided JSON file (the storyboard) and interpolates it directly into the prompt sent to the LLM. If the storyboard file contains malicious instructions, the model might follow them instead of performing the transition design.
- Ingestion point: The
scenesdata from the file path provided in the--storyboardargument inscripts/generate_transitions.py. - Boundary markers: The script does not use delimiters (like XML tags or clear separators) or specific instructions to the model to treat the storyboard content as untrusted data.
- Capability inventory: The script performs file system writes to the path specified by the
--outputargument. - Sanitization: No sanitization or validation is performed on the strings within the storyboard JSON.
- [EXTERNAL_DOWNLOADS]: The Python script requires the
volcenginesdkarkruntimepackage to interact with the ByteDance Volcengine Ark API. This is a dependency for a well-known cloud service provider.
Audit Metadata