The Agent Skills Directory

[SAFE]: The skill is designed for system diagnostics and health checks. It follows security best practices by defaulting to read-only operations and explicitly instructing the agent to avoid dumping sensitive information such as tokens, API keys, or full configuration file contents.
[COMMAND_EXECUTION]: The diagnostic script executes several local commands including cmux, stat, grep, and wc. These are standard tools used for checking system state, file metadata, and the presence of specific configuration markers without exposing the raw data itself.
[DATA_EXFILTRATION]: While the skill accesses various configuration paths (e.g., ~/.config/cmux/, ~/.codex/hooks.json), it only reports on file existence, size, modification times, and the presence of specific strings. It implements a safe_path function to redact the local user's home directory path from reports, reducing metadata leakage.

cmux-diagnostics