miniprogram-development
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill includes a code example for AI model integration that takes user-provided natural language and interpolates it into a prompt message. * Ingestion points: The
userInputvariable in the AI invocation script inSKILL.md. * Boundary markers: The example uses a system role prompt but lacks explicit delimiters or escaping for the user message content. * Capability inventory: The example code logs the model output to the console, which is a low-risk capability. * Sanitization: The snippet does not include input validation or output sanitization. - [COMMAND_EXECUTION]: The skill provides pre-defined CLI commands for opening the WeChat Developer Tools. * Evidence: Includes specific file paths for the WeChat Developer Tools executable on Windows and macOS to facilitate opening project directories via the
cli.batandclitools. - [EXTERNAL_DOWNLOADS]: Recommends using Icons8 for fetching resource icons. * Evidence: Provides a specific URL format for
img.icons8.com, which is a well-known service for design assets.
Audit Metadata