RunComfy Agent Skills - Setup Guide¶
Source: prime-skills/runcomfy-agent-skills (11.5M+ combined installs, Oct 8, 2026 snapshot) GitHub: prime-skills/runcomfy-agent-skills Platform: RunComfy - cloud GPU platform for AI media generation Category: AI Media / Video Production Quality Tier: 🟡 Beta (first seen Jul 13, 2026)
RunComfy Agent Skills lets Hermes agents access 30+ AI models for video generation, image creation, avatar synthesis, and music production through a single CLI. For Hermes agents producing UGC videos, social content, or AI-generated media, RunComfy provides a complementary backend to HyperFrames and an alternative to HeyGen.
Installation¶
# Full publisher install (all 30 skills)
npx skills add prime-skills/runcomfy-agent-skills
# Or individual skills
npx skills add prime-skills/runcomfy-agent-skills --skill ai-video-generation
npx skills add prime-skills/runcomfy-agent-skills --skill video-edit
Prerequisites¶
| Requirement | Details |
|---|---|
| RunComfy account | Sign up at runcomfy.com. Free tier available. |
| RunComfy CLI | npm i -g @runcomfy/cli then runcomfy login |
| API token (CI) | Set RUNCOMFY_TOKEN=<token> for headless/agent use |
| Hermes Agent | Any version with skills support |
Core Skills¶
Video Generation¶
| Skill | Installs | Purpose |
|---|---|---|
| video-edit | 431.0K | Intent-routed video editing - restyle, motion transfer, outfit/background swap |
| image-to-video | 429.4K | Animate still images - HappyHorse I2V, Wan 2.7, Seedance 2.0 |
| ai-video-generation | 374.4K | Full text-to-video + image-to-video via single CLI |
| ai-avatar-video | 371.4K | Talking head / avatar video with lip-sync |
| video-inpainting | 369.5K | Remove objects/people from video |
| video-outpainting | 367.9K | Extend video frame boundaries |
| video-extend | 368.4K | Extend video duration - Veo-style, Kling, Seedance |
| lipsync | 368.9K | Audio-driven lip synchronization |
Image Generation & Editing¶
| Skill | Installs | Purpose |
|---|---|---|
| ai-image-generation | 374.4K | Full RunComfy image-model catalog |
| image-edit | 428.7K | Intent-routed image editing |
| gpt-image-2 | 75.8K | OpenAI GPT image generation |
| gpt-image-edit | 427.7K | OpenAI GPT image editing |
| nano-banana-2 | 429.1K | Nano Banana 2 image generation |
| flux-2-klein | 427.5K | Flux 2 Klein generation |
| flux-kontext | 428.5K | Flux Kontext contextual generation |
| controlnet-pose | 369.0K | Pose-guided generation |
Audio, Music & Effects¶
| Skill | Installs | Purpose |
|---|---|---|
| ai-music | 360.5K | AI music generation |
| elevenlabs-music-generation | 368.4K | ElevenLabs music via RunComfy |
| relight | 368.1K | Professional scene relighting |
| face-swap | 371.3K | Face swapping with identity preservation |
Full list: 30 skills including seedance-v2, wan-2-7, happyhorse-1-0, kling-3-0, ace-step, codex-pet, runcomfy-cli.
CorpusIQ Use Cases¶
- Daily UGC Video Pipeline - RunComfy as HeyGen alternative for avatar video; HyperFrames for composition
- Social Media Content - Generate 60-second product demos from script + image
- Brand Assets - Programmatic image generation for posts, headers, ads
- Multi-Model Routing - Intent-based model selection: "animate this" picks best model automatically
Configuration¶
npm i -g @runcomfy/cli
runcomfy login
# or: export RUNCOMFY_TOKEN=<token>
npx skills add prime-skills/runcomfy-agent-skills
Mirror Listings (Oct 8, 2026 sweep)¶
This guide covers the prime-skills/runcomfy-agent-skills listing, which carries the install base. Related redistribution sources observed: gencraft-labs/skills republishes a partial RunComfy set (5 indexed listings, ~80K combined, first seen Oct 6, 2026; its README defers installs to agentspace-so/runcomfy-agent-skills). Treat other RunComfy skill sources as mirrors of this one.
Related Skills¶
- HyperFrames - Hermes-native video composition (complementary)
- HeyGen Video Automation - Avatar video (RunComfy is an alternative)
- media-use - Asset resolution layer for video projects
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.