mockup-sandbox
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests existing application source code (via the companion skill mockup-extract) and user prompts to generate new UI mockups. This establishes an attack surface where malicious instructions in the input data could influence the agent's code generation. The instructions do not specify sanitization or clear boundary markers for this untrusted input.
- Ingestion points: existing component extraction process.
- Boundary markers: None specified in the instructions.
- Capability inventory: File writing to the sandbox directory, Vite server execution, and package installation via pnpm.
- Sanitization: No sanitization steps are documented for processing extracted code.
- [DYNAMIC_EXECUTION]: The skill's primary function is to generate .tsx component files and serve them using a Vite development server. This involves runtime code generation and execution. Additionally, a custom Vite plugin (mockupPreviewPlugin.ts) dynamically generates a component registry file at runtime.
- [COMMAND_EXECUTION]: The skill requires the execution of shell commands for environment setup and management, specifically instructing the use of pnpm install from the project root and workflow restarts to manage the Vite development server.
- [EXTERNAL_DOWNLOADS]: The skill fetches external fonts from Google Fonts, a well-known service, via HTML link tags. It also facilitates the installation of third-party dependencies from standard registries using pnpm.
Audit Metadata