The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill provides implementation patterns for secure credential storage, prioritizing the OS Keyring (via the keyring library) and environment variables over plain-text storage. It explicitly advises against committing secrets to version control.
[PRIVILEGE_ESCALATION]: The provided Python templates include security-conscious file handling, such as using chmod 0o600 to restrict access permissions for sensitive configuration and environment files.
[DATA_EXFILTRATION]: The skill demonstrates standard API interaction patterns using the httpx library targeting a placeholder domain (api.example.com). No suspicious or unauthorized data transmission behaviors were identified.
[INDIRECT_PROMPT_INJECTION]: The skill identifies that CLI tools act as ingestion points for untrusted data via flags and arguments. It mitigates this risk by recommending structured data formats (JSON) and utilizing robust argument parsing libraries (Typer/Click) for validation.

cli-ops