The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Node.js scripts and CLI tools such as claude, maestro, simctl, and adb using spawnSync and spawn. These executions are central to the skill's purpose of validating and testing other agent skills within a project.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection by processing external data, such as routing evaluation prompts and the content of other skills being validated. The implementation follows best practices for tool use within a developer-controlled environment, and the risk is assessed as low.
[SAFE]: All file system interactions, such as reading configuration files like package.json or tsconfig.json and writing scaffolded skill files, are performed using standard Node.js modules and are limited to the project directory or temporary folders.
[SAFE]: Instructions for referencing external resources point to well-known technology domains, such as GitHub and official Expo repositories.

skill-maintenance