kling-3-0
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
runcomfyCLI to communicate with the RunComfy Model API to generate video content.\n - Evidence: Execution of
runcomfy run kling/kling-3.0/<tier>/<mode>with JSON-formatted input parameters.\n- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the official@runcomfy/clipackage from the public npm registry.\n - Evidence: Prerequisites section specifies
npm i -g @runcomfy/cli.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes user-supplied text prompts and image URLs.\n - Ingestion points: The
prompt,image_url, andtail_image_urlfields in the input schema defined inSKILL.md.\n - Boundary markers: User input is encapsulated within a JSON object passed to the CLI's
--inputargument.\n - Capability inventory: The skill executes the
runcomfyCLI, which performs network operations to the model API and downloads resulting video files.\n - Sanitization: The skill documentation states that the CLI does not perform shell expansion on the JSON input, which mitigates the risk of command injection via user-supplied prompts.
Audit Metadata