Image & Video for AI Agents

This skill provides access to the Wan 2.7 video generation model via the RunComfy API. It supports text-to-video generation with features including audio-driven lip-sync, multi-reference motion control, and configurable output settings.

This skill routes video editing tasks to the appropriate RunComfy model based on user intent. It supports restyling, motion transfer, and identity-stable edits by selecting the best model for the specific request.

This skill provides access to the OpenAI GPT Image 2 edit endpoint via the RunComfy CLI. It is designed for tasks requiring high-fidelity image modifications, such as multilingual text replacement, layout adjustments, and multi-image composition while preserving subject identity.

This skill provides precise, single-reference image editing using the Flux 1 Kontext Pro model via the RunComfy API. It is designed for tasks requiring high-fidelity preservation of source identity while applying specific, declarative changes.

This skill provides access to the Flux 2 Klein image generation models via the RunComfy API. It supports both the 4B variant for rapid, sub-second concepting and the 9B variant for high-fidelity final outputs.

Seedance 2.0 Pro is a multimodal video generation skill that creates cinematic clips with native lip-synced audio. It allows users to combine character images, reference videos, and audio files to maintain consistent identity and narrative across generated content.

HappyHorse 1.0 is a text-to-video model hosted on the RunComfy Model API that generates 1080p video with synchronized audio. It is designed for multi-shot storytelling and maintaining character consistency across generated clips.

This skill routes image-to-video requests to the appropriate model based on user intent. It supports portrait animation, custom voiceover lip-sync, and multi-modal composition using the RunComfy catalog.

This skill routes image editing tasks to the appropriate model within the RunComfy catalog based on user intent. It supports batch processing, multilingual text rewriting, precise local edits, and mask-driven inpainting.

This skill provides an image-to-image editing interface for the Google Nano Banana 2 model via the RunComfy API. It supports batch processing of up to 20 images and allows for precise spatial edits while preserving subject identity.