shader-liquid-glass
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The font_to_glsl.py script downloads font assets from the official Google Fonts repository on GitHub. This is a legitimate operation from a well-known source, providing required data for the skill's typography features.
- [COMMAND_EXECUTION]: The skill executes the curl and sips utilities via Python's subprocess.run in font_to_glsl.py. These calls are implemented using list-based arguments, which prevents shell command injection and ensures that user-provided inputs are treated strictly as data.
- [INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided text and image paths to generate shaders, creating an attack surface for indirect injection.
- Ingestion points: Input provided to the font_to_glsl.py script via the --text and --image flags.
- Boundary markers: User text is embedded inside GLSL line comments in the generated output files.
- Capability inventory: The skill has the capability to write files, fetch data from the internet via curl, and process images using system tools in font_to_glsl.py.
- Sanitization: The script utilizes HTML entity escaping for characters in the title field of preview pages. While text in GLSL comments is not escaped, the potential risk is limited to disrupting the syntax of the generated shader output rather than influencing the agent's internal logic.
- [SAFE]: No evidence of malicious behavior, such as credential exfiltration, obfuscation, or persistence, was found. The skill's operations are transparent and consistent with its purpose as a creative coding assistant.
Audit Metadata