The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the runcomfy CLI tool using Bash, as restricted in the allowed-tools frontmatter. This execution is limited to the skill's primary functionality and does not exhibit signs of arbitrary command execution.- [EXTERNAL_DOWNLOADS]: The skill utilizes npm (a well-known service) to install its required @runcomfy/cli package. These downloads are directed to official package registries and are considered safe under standard security evaluations.- [CREDENTIALS_UNSAFE]: The skill includes instructions for managing API tokens using environment variables or localized configuration files. It provides explicit security warnings against exposing these tokens in logs or prompts, demonstrating a security-first approach to credential handling.- [SAFE]: The skill identifies the potential for indirect prompt injection via user-provided image or audio URLs and provides sensible mitigation strategies, such as cautioning that external assets should be considered untrusted content.

ai-video-generation