The Agent Skills Directory

[COMMAND_EXECUTION]: The main execution script scripts/run-matrix.py uses subprocess.run to interface with the container CLI for building images and running containers. This is the primary intended function of the skill.
[SAFE]: The skill implements a sophisticated secret redaction mechanism in scripts/run-matrix.py. It automatically identifies values of sensitive environment variables passed via inheritEnv and ensures they are deleted from stdout, stderr, and all generated evidence artifacts before they are saved to disk or included in reports. It also uses secure file permissions (0700 for directories, 0600 for files) and atomic write operations to protect the integrity of the results.
[EXTERNAL_DOWNLOADS]: The assets/agent-cli-Containerfile downloads official packages from the well-known NPM registry, including @anthropic-ai/claude-code, @openai/codex, and typescript. These dependencies are pinned to specific versions to ensure consistency and security.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of technical validation harnesses inside isolated Apple Containers. While it executes shell commands (e.g., via sh -lc in the manifest), this execution is scoped within the container environment, providing a significant layer of isolation from the host system.

apple-container-validation