carousel-design
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to manage its environment, including
npm install puppeteerfor dependency management and running generated Node.js scripts. - [DYNAMIC_EXECUTION]: The skill uses a template-based approach to generate a JavaScript build script at runtime. This script incorporates user-provided slide data and configuration values before being executed via Node.js to perform the rendering tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill processes slide content (such as headlines and terminal text) by interpolating it directly into HTML templates. This creates a potential attack surface where malicious input could attempt to execute scripts within the Puppeteer headless browser environment.
- Ingestion points: Slide data provided in JSON format, typically from the
/carousel-copycommand output. - Boundary markers: None identified; data is directly replaced into template placeholders.
- Capability inventory: File system access (
fs.writeFileSync,fs.readFileSync) and headless browser operations (puppeteer.launch,page.goto,page.screenshot). - Sanitization: No explicit sanitization or escaping of the HTML-injected content is mentioned in the rendering logic.
Audit Metadata