seedance-2-5
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted external media data, creating a potential surface for indirect prompt injection.
- Ingestion points: The skill processes external assets including up to 30 images, 10 videos, and 10 audio clips via provider-specific URLs (e.g.,
image_urls,video_urls) as described inSKILL.md. - Boundary markers: There are no explicit instructions or delimiters provided to the agent to ignore or isolate instructions that might be embedded within the metadata or content of the external reference assets.
- Capability inventory: The skill grants the agent the ability to call external video generation tools (
seedance_video,seedance_ark,runway_video,comfyui_video) which ingest and process these external inputs. - Sanitization: The skill does not specify any sanitization, validation, or filtering mechanisms for the data retrieved from the external URLs before it is passed to the generation models.
- [NO_CODE]: The skill consists entirely of markdown instructions and metadata, with no accompanying scripts or binary files, reducing the risk of direct malicious code execution.
Audit Metadata