The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes dynamic context injection (shell commands executed at load time via the !command syntax) within its scaffolding scripts located in the scripts/ directory. These commands are used to auto-detect project context, such as identifying the active test framework in package.json, searching for existing fixtures or handlers, and checking the Python version. These operations are performed using standard utilities like grep, find, and wc, and do not inject user-supplied arguments into the shell execution, making the implementation safe.
[EXTERNAL_DOWNLOADS]: References to external resources are limited to official documentation (mswjs.io, GitHub), common testing libraries, and placeholder examples for API endpoints. There are no instructions for downloading or executing scripts from untrusted remote sources.
[DATA_EXFILTRATION]: The skill does not contain any patterns for accessing or transmitting sensitive data. It explicitly promotes security best practices by including rules and checklists for filtering sensitive information, such as authorization headers and API keys, from test recordings (VCR cassettes).

testing-unit