The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the platform's native shell execution feature (!\command`) used to gather context and automate tasks. The documentation includes guidelines on safe usage and restricting tool access via the allowed-tools` frontmatter field.
[REMOTE_CODE_EXECUTION]: Practical examples demonstrate integration with standard developer tools and services such as Git, NPM, Kubernetes, and the GitHub CLI. These are documented as part of intended command functionality for developers.
[PRIVILEGE_ESCALATION]: Documentation regarding testing and debugging includes benign examples of using system utilities like dd, rm, and chmod for managing temporary test files and verifying security restrictions.
[INDIRECT_PROMPT_INJECTION]: The skill describes patterns for processing external data (file contents and user arguments). It proactively provides mitigation strategies through 'Validation Patterns' that teach developers how to sanitize and verify inputs before execution.

command-development