shot-quality-gate
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
cinefoundrycommand-line tool to perform quality analysis. It executes shell commands passing project IDs and file paths as arguments to evaluate image frames. - [INDIRECT_PROMPT_INJECTION]: The skill processes AI-generated images through a vision model, which creates a vulnerability to indirect prompt injection if an image contains adversarial visual elements.
- Ingestion points: The skill ingests generated frames and reference images (
.png) from the local file system (e.g.,projects/<id>/frames/) into the vision model context. - Boundary markers: The workflow lacks explicit delimiters or instructions for the vision model to ignore or differentiate between instructional content and the visual data being analyzed.
- Capability inventory: The skill has the capability to execute subprocess commands (
cinefoundry) and automate prompt modifications for regeneration based on the vision model's output. - Sanitization: There is no evidence of visual sanitization or filtering applied to the generated frames before they are processed by the quality gate model.
Audit Metadata