image-to-svg
Warn
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill includes a C source file (scripts/nn_assign.c) that is automatically compiled into a binary and executed at runtime by the Python pipeline (scripts/pipeline.py) to accelerate nearest-neighbor calculations. This dynamic compilation and execution of binary code is a potential risk vector if the source file were to be tampered with.
- [COMMAND_EXECUTION]: The script uses the subprocess.run function to call system utilities, including the gcc compiler, the compiled nn_assign binary, and ImageMagick (convert or magick) for image smoothing. While arguments are passed as a list to reduce command injection risk, it introduces a dependency on the security and availability of these system-level tools.
- [EXTERNAL_DOWNLOADS]: The documentation (SKILL.md) provides instructions to install standard Python libraries (opencv-python-headless, scikit-image, scipy, scikit-learn) and system utilities (librsvg2-bin) from official registries.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted image data through high-capability tools like ImageMagick and custom compiled binaries, which represents a potential attack surface.
- Ingestion points: Image files provided via the source_path parameter in scripts/pipeline.py.
- Boundary markers: No specific boundary markers or 'ignore' instructions are used for the image data content.
- Capability inventory: Execution of compiled C binaries and ImageMagick via subprocess.run calls in scripts/pipeline.py.
- Sanitization: Input paths and parameters are used within subprocess lists, though the binary image data content itself is not sanitized.
Audit Metadata