blender-render
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the agent context (such as file paths, camera names, and scene identifiers) and processes it through Blender's rendering engine and file system operations.
- Ingestion points: Multiple tools accept user-defined strings for filepath, camera_names, and output_pattern parameters.
- Boundary markers: No explicit delimiters are used in the command arguments, though the documentation states that background workers are configured to disable automatic scripts.
- Capability inventory: The skill can perform file system writes and execute background processes via the Blender API.
- Sanitization: The skill employs secure file handling practices in capture_viewport.py and relies on the Blender API for internal parameter validation.
Audit Metadata