code-to-image
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references external resources to style the generated images, including the Tailwind CSS runtime from
cdn.tailwindcss.comand typography from Google Fonts (fonts.googleapis.com). These are well-known and expected services for the stated purpose. - [COMMAND_EXECUTION]: Instructs the agent to use the
curlcommand-line tool to interact with the conversion API endpoint. - [DATA_EXFILTRATION]: User-provided code snippets are transmitted to the
html2png.devAPI endpoint. This service belongs to the skill's author and the transmission is essential for the skill's primary function of image generation. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code snippets from users to be processed by an external API.
- Ingestion points: User-supplied code blocks in SKILL.md examples or provided during execution.
- Boundary markers: No explicit markers or delimiters are provided to differentiate instructions from data in the transmitted payload.
- Capability inventory: The skill utilizes network capabilities (
curl) to communicate with thehtml2png.devservice. - Sanitization: There is no specific evidence of sanitization for the code content before it is converted to HTML.
Audit Metadata