modern-frontend-design
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to take user product concepts and generate frontend code, then instructs the agent to verify this code using shell commands. This creates an attack surface where a malicious prompt could trick the agent into generating harmful code (e.g., in a package.json postinstall script) which is then executed during the 'Final Automated Testing' step. * Ingestion points: User product concepts and design requests processed in SKILL.md. * Boundary markers: None present. * Capability inventory: Includes 'npm install', 'npm run build', 'npx next lint', 'npx tsc', and local server testing via 'curl'. * Sanitization: None mentioned.
- [COMMAND_EXECUTION]: Step 11 explicitly commands the agent to execute a series of shell operations in the user's project environment, including installation of dependencies, compilation, and running a development server to verify the build.
- [REMOTE_CODE_EXECUTION]: The instruction to run 'npm install' triggers the download and execution of external packages from the NPM registry. The specific packages executed are determined by the dynamically generated 'package.json' file, which is based on user input.
- [DYNAMIC_EXECUTION]: The verification process involves 'npm run build' and 'npm run dev', which execute scripts defined within the generated project files at runtime.
Audit Metadata