The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands including docker ps to verify container status and curl to check local health endpoints. These operations are performed as part of the primary diagnostic function and are restricted to the local system.
[DATA_EXFILTRATION]: The skill references sensitive local file paths like .env and .env.local to check for their existence. This is a local check used to ensure the environment is configured correctly; no sensitive data is read, processed, or exfiltrated.
[REMOTE_CODE_EXECUTION]: All scripts and logic are local to the skill package. No external scripts are downloaded or executed from remote sources, and dynamic logic is limited to local configuration resolution.
[SAFE]: The configuration resolution process in resolve_sanity.py uses shell-safe quoting via the shlex library to ensure that data loaded from local configuration files cannot be used for command injection when processed by the shell.

dev-sanity