patrick-visual-language
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/render.pyscript usessubprocess.runto launch local browser binaries (Google Chrome or Chromium) and thesipsimage tool. This is a functional requirement to generate screenshots and measure the mobile viewport layout of the HTML files styled using this skill. - [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing untrusted data as it is designed to audit and render HTML files that may be generated by the agent or provided by external sources.
- Ingestion points: HTML file content is ingested by
scripts/audit.py(viaPath.read_text) andscripts/render.py(via file loading in Chrome). - Boundary markers: No specific boundary markers are used for the HTML content as it is the primary artifact being processed.
- Capability inventory: The skill includes file system read access and the ability to execute external rendering tools via the command line.
- Sanitization: The
audit.pyscript performs static analysis and issues warnings if external network references are found, which helps prevent unintended data exposure or external dependencies.
Audit Metadata