create-workflow-diagram
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes several shell commands to function correctly, including 'npm install' for dependencies, 'npx playwright install chromium' for the browser setup, and 'node screenshot-diagram.js' to capture the final diagram. This is expected behavior for the tool's primary purpose of exporting images.
- [EXTERNAL_DOWNLOADS]: The skill installs the 'playwright' library from the official NPM registry and downloads a Chromium browser instance. Playwright is a well-known and established automation service maintained by Microsoft.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection and path traversal as it processes untrusted user input to generate file paths and HTML content.
- Ingestion points: User-provided workflow descriptions and diagram names are captured in 'SKILL.md' (Step 1.1).
- Boundary markers: Absent. There are no instructions for the agent to ignore or delimit potentially malicious instructions within the workflow description.
- Capability inventory: The skill can create directories, write HTML files ('diagram.html'), and execute shell commands ('node screenshot-diagram.js').
- Sanitization: Absent. There is no explicit requirement to sanitize user-provided diagram names or workflow labels, which could lead to path traversal or script injection in the generated HTML.
Audit Metadata