kling-3-0
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to install the
@runcomfy/clipackage from the official npm registry to enable model interaction and management. - [COMMAND_EXECUTION]: The skill utilizes the
runcomfyCLI tool to submit generation requests and manage output. The tool uses JSON formatting for input parameters to reduce the risk of shell injection. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text and external image URLs, which are then processed by the Kling 3.0 video model.
- Ingestion points: User inputs for
prompt,image_url, andtail_image_urldefined in the input schema withinSKILL.md. - Boundary markers: The skill does not implement explicit delimiters or markers to isolate user-provided prompts within the model execution context.
- Capability inventory: The skill can execute local shell commands via the
runcomfytool and perform file writes to the directory specified by the--output-dirargument. - Sanitization: The documentation specifies that the CLI avoids shell expansion for inputs and explicitly notes the potential for image-based prompt injection at the model level.
Audit Metadata