The Agent Skills Directory

[SAFE]: The skill provides utility scripts and reference documentation for context engineering and prompt optimization. The primary logic in scripts/context_audit.py involves lexical analysis and reporting on markdown structures.
[COMMAND_EXECUTION]: The skill's integration tests (tests/test_cli_integration.py) utilize subprocess.run() to invoke the local audit script. This implementation is standard for CLI tool validation and is restricted to executing the skill's own code within a test environment.
[EXTERNAL_DOWNLOADS]: The skill identifies standard development dependencies such as pytest and tiktoken in its tests/requirements-dev.txt. These are recognized packages from established registries.
[SAFE]: The skill contains no patterns related to prompt injection, credential exfiltration, persistence, or obfuscation. All file system operations are scoped to auditing the specific directories and files provided as inputs by the user.

unhobble