ai-video-generation
๐ฏSkillfrom doany-ai/skills
|Generate AI videos on RunComfy through a smart router across the full video-model catalog including HappyHorse 1.0, Wan 2-7, Seedance v2, Kling 3.0, Veo 3-1, and Hailuo 2-3. Covers text-to-video, image-to-video, and video-extend, automatically selecting the best model for the user's intent.
Overview
AI Video Generation is a Claude Code skill that smart-routes video creation across the full RunComfy video-model catalog, supporting text-to-video, image-to-video, and video-extend workflows. It selects the best model for the user's intent from options including HappyHorse 1.0 (Arena #1 with native in-pass audio), Wan 2-7 (open weights with audio-driven lip-sync), Seedance v2 (multi-modal cinematic with up to 9 reference images), Kling 3.0 (4K multi-shot character identity), Veo 3-1 (physics-respecting motion), and Hailuo 2-3 (natural motion for real-world subjects). Each model ships with documented prompting patterns and exact CLI invocations.
Key Features
- Intent-based model routing: Automatically classifies whether the user needs general-purpose video, audio-driven lip-sync, cinematic multi-reference composition, 4K hero shots, physics-accurate motion, or natural real-world animation, and picks the matching model
- Native audio generation: HappyHorse 1.0 generates synchronized audio in-pass (describe audio inline with the prompt), while Wan 2-7 supports audio-driven lip-sync from a specific voiceover MP3
- Multi-modal conditioning: Seedance v2 Pro accepts up to 9 reference images, 3 reference videos, and 3 reference audio tracks for complex cinematic compositions with lens and film language support
- Text-to-video, image-to-video, and video-extend: Full coverage from prompt-only generation through still-image animation to extending existing clips via Veo 3-1's extend endpoint
- Quality-cost tiers per model: Multiple tiers available (e.g., Kling 3.0 4K/Pro/Standard) so users can iterate cheaply and upgrade to production quality for final delivery
Who is this for?
This skill is for developers, content creators, and production teams who need to generate AI videos through Claude Code without manually comparing video models. It serves social media creators producing vertical clips with audio, brand teams creating cinematic ad frames, production pipelines needing dialog lip-sync from voiceover files, and anyone who wants physics-accurate product spins or multi-shot character narratives through a single CLI command.
Same repository
doany-ai/skills(29 items)
Installation
npx vibeindex add doany-ai/skills --skill ai-video-generationnpx skills add doany-ai/skills --skill ai-video-generation~/.claude/skills/ai-video-generation/SKILL.mdSKILL.md
More from this repository10
A Claude Code skill for image outpainting on RunComfy that extends images beyond their original canvas, supporting aspect ratio changes, uncropping, and canvas expansion by routing across Nano Banana 2 Edit, GPT Image 2 Edit, and FLUX Kontext Pro.
A Claude Code skill for pose-conditioned image and video generation on RunComfy, routing across Kling 2-6 Motion Control (video motion transfer), Wan 2-2 Animate (audio-driven character animation), and Z-Image Turbo ControlNet LoRA (pose-conditioned image generation).
A Claude Code skill that creates AI avatar and talking-head videos on RunComfy, intelligently routing across OmniHuman, Wan 2-7, HappyHorse, and Seedance models based on user intent such as UGC voiceover, virtual presenter, or lip-synced character.
A Claude Code skill for generating images with Google Nano Banana 2, a Gemini-family flash-tier text-to-image model hosted on RunComfy, optimized for rapid iteration, social thumbnails, and in-image typography rendering.
Extend a video's spatial canvas on RunComfy โ uncrop, change aspect ratio (e.g., 9:16 to 16:9), or add environment beyond the original frame while preserving the central action. Routes through Wan 2-7 edit-video and dedicated ComfyUI outpaint workflows.
A Claude Code skill that extends existing video clips on RunComfy using Google Veo 3-1's extend-video endpoints, continuing clips past their duration cap or chaining narrative shots while preserving consistent motion, lighting, and subject identity.
Generate full songs and instrumental tracks with ElevenLabs Music on RunComfy โ studio-quality 44.1 kHz stereo audio from 5 seconds to 5 minutes with section-level control (Intro, Verse, Chorus, Bridge), multilingual vocals, and commercial-friendly output.
A Claude Code skill for generating and editing images with OpenAI GPT Image 2 (ChatGPT Images 2.0) hosted on RunComfy, with strengths in embedded text, logos, multilingual typography, and precise multi-element prompt following.
Animate any still image into video on RunComfy, routing to the best i2v model for each intent โ HappyHorse 1.0 for general animations with native audio, Wan 2.7 with audio_url for custom-voiceover lip-sync, or Seedance 2.0 Pro for multi-modal animation from image plus reference video and audio.
A Claude Code skill that acts as a smart router for video editing on RunComfy, automatically selecting the best model -- Wan 2.7 Edit-Video for general restyle, Kling 2.6 Pro for motion transfer, or Lucy Edit Restyle for lightweight outfit/background swaps.