The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the runcomfy CLI to execute various image-to-video models. This is the core functionality of the skill and is performed through a documented command-line interface.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @runcomfy/cli package from the npm registry and references a GitHub repository (agentspace-so/runcomfy-skills) for skill management. These are standard external resources for the vendor's ecosystem.
[PROMPT_INJECTION]: The skill processes user-provided inputs such as image_url, audio_url, and prompt strings.
Ingestion points: User-supplied URLs and text prompts defined in the Invoke sections of SKILL.md.
Boundary markers: Data is passed to the CLI using a JSON-formatted string within the --input flag, which serves as a delimiter.
Capability inventory: The skill executes the runcomfy command to interact with remote APIs and downloads resulting media files.
Sanitization: The documentation explicitly states that the CLI does not shell-expand the prompt, mitigating traditional command injection while acknowledging the inherent risk of model-level prompt injection from external content.

image-to-video