The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches skill definitions and instructions from the agentspace-so/runcomfy-skills repository on GitHub and suggests installing the @runcomfy/cli package from the official npm registry.
[COMMAND_EXECUTION]: Executes the runcomfy CLI tool to interact with the RunComfy Model API. The skill explicitly notes that user-provided prompts are passed as JSON strings to avoid shell injection vulnerabilities.
[DATA_EXPOSURE]: Documents the location of the API token at ~/.config/runcomfy/token.json. It notes that the CLI manages this file with restricted permissions (0600) to ensure security.
[PROMPT_INJECTION]: While the skill ingests user prompts and image URLs, it implements strong boundary markers by passing data through a structured JSON body to the CLI, mitigating direct and indirect injection risks.

flux-kontext