chart-craft-plus
Warn
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructions specify running
npx getdesign@latest add <slug>to fetch brand design tokens. This command downloads and executes the latest version of an external package from the npm registry at runtime, which represents a remote code execution vector from an unverified source. - [COMMAND_EXECUTION]: The skill employs several shell commands to manage files and preview output, including
mkdir -pto create local directories,cdto change the working directory, andopento launch the generated HTML file in the user's default browser. - [EXTERNAL_DOWNLOADS]: The use of
npxinvolves downloading external assets or packages from a remote repository during the 'Brand Style Extension' workflow, which occurs outside of the skill's local files. - [PROMPT_INJECTION]: The skill processes untrusted user input (descriptions and chart data) to generate HTML and CSS. While instructions advise the agent not to use Javascript, the lack of explicit sanitization or boundary markers (such as 'ignore embedded instructions' warnings) for the interpolated user content creates a surface for indirect prompt injection or cross-site scripting (XSS) in the generated file.
Audit Metadata