review-annotations
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts (
scripts/multi_pass_annotate.pyandscripts/draw_annotations.py) to automate the annotation process. These scripts are run with arguments derived from the project configuration and user-provided image paths. - [INDIRECT_PROMPT_INJECTION]: The skill processes external images and YAML configuration files, creating a potential surface for indirect prompt injection if the vision model or scripts ingest untrusted instructions embedded in image data or metadata.
- Ingestion points: Processes image batches and
yolo-project.yamlas described in Workflow Step 3. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the image analysis pass.
- Capability inventory: Executes command-line scripts and performs multimodal vision analysis.
- Sanitization: The instructions do not specify validation or sanitization of image content or annotation text before processing.
Audit Metadata