The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and installs the @runcomfy/cli package from the npm registry to enable video processing capabilities.
[EXTERNAL_DOWNLOADS]: Downloads video and image assets from external, user-provided URLs for processing by the RunComfy models.
[COMMAND_EXECUTION]: Invokes the runcomfy CLI tool to perform video transformations, passing parameters through a JSON-formatted input string.
[CREDENTIALS_UNSAFE]: References the use of API tokens stored in ~/.config/runcomfy/token.json or the RUNCOMFY_TOKEN environment variable for service authentication.
[PROMPT_INJECTION]:
Ingestion points: User-supplied natural language prompts and media content from external URLs (found in SKILL.md).
Boundary markers: Commands use JSON encapsulation for inputs; the documentation states the CLI does not perform shell expansion on these prompts.
Capability inventory: Ability to execute the local runcomfy binary and write output files to a specified directory.
Sanitization: Relies on the RunComfy Model API's internal filters; documentation explicitly warns about the risks of image-based prompt injection from untrusted URLs.

video-edit