infographic-generator
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data retrieved from external sources through research tools and incorporates it into an HTML layout. This layout is subsequently loaded and rendered by a headless Playwright browser. Malicious HTML or scripts present in the research data could potentially be executed within the browser context.\n
- Ingestion points: Research data from tavily_search and tavily_research (SKILL.md, Step 0).\n
- Boundary markers: No delimiters or warnings identified in the HTML construction phase to isolate untrusted content.\n
- Capability inventory: Execution of Node.js scripts, file system writes, and network operations via Tavily tools.\n
- Sanitization: The instructions do not describe a process for sanitizing or escaping the researched content before it is inserted into the HTML layout.\n- [COMMAND_EXECUTION]: The export process uses shell commands that incorporate a user-defined topic to construct file paths. If the topic string contains shell metacharacters and is not correctly escaped when the agent generates the script, it could lead to command injection.\n- [DYNAMIC_EXECUTION]: The skill dynamically generates an HTML file and uses node -e to execute a script that renders this file to a PNG image.\n- [EXTERNAL_DOWNLOADS]: The skill downloads the playwright package from the NPM registry to provide image export functionality.
Audit Metadata