Search, download, and install Civitai models into ComfyUI for local generation.
npx skills add https://github.com/artokun/comfyui-mcp --skill civitai
Please help me install this Agent Skill. Skill: civitai Source: https://github.com/artokun/comfyui-mcp/tree/main/plugin/skills/civitai Command: npx skills add https://github.com/artokun/comfyui-mcp --skill civitai
Finding the right checkpoint, LoRA, or embedding on Civitai and getting it working inside ComfyUI normally requires manual browsing, downloading, file placement, and figuring out trigger words and base-model compatibility. This Skill automates that entire find-install-generate loop through comfyui-mcp's built-in Civitai actions.
download_model with action:"search_civitai", no API key required.model_version_id directly into the correct ComfyUI subfolder (loras, checkpoints, vae, controlnet, embeddings) with action:"download_civitai".generate_image.loras, wires it into the graph, and inserts its trigger words into the prompt.Ask the agent to find an anime-style LoRA for Flux on Civitai, install it into ComfyUI, and generate an image using its trigger words.
High-intent search queries and answers about installing and using this skill.
Use download_model with action "search_civitai" and a query, filtering by types and base_models. Then call action "download_civitai" with the chosen model_version_id and the correct target_subfolder such as loras or checkpoints.
Parse the modelVersionId from the pasted URL and pass it to download_civitai, or pass the model id to resolve the latest version. Raw civitai.com/api/download links also work via action "download" with CIVITAI_API_TOKEN set.
No API key is needed for search and standard downloads. CIVITAI_API_TOKEN from civitai.com/user/account unlocks gated and early-access downloads and gated search results, set once in panel Settings or as an environment variable.
LoRAs are base-model specific, so a Flux LoRA will not load on an SDXL checkpoint. Always filter searches with the base_models parameter matching your checkpoint family, such as Flux.1 D, SDXL 1.0, or SD 1.5.
The built-in search_civitai action covers discovery, download, and local generation with no extra server. The official Civitai MCP at mcp.civitai.com adds community features like browsing example images, collections, posting, and reviews.
timoncoolIntegrates Meilisearch with Civitai to search models, browse images, and download LoRAs/Checkpoints.
humanerd-drewGenerate images, video, and audio via ComfyUI REST and WebSocket APIs.
artokunDetail ComfyUI model compatibility requirements for SD 1.5, SDXL, Flux, SD3, and LTXV.
RachasumanthPackage trained AI models and publish them to Hugging Face Hub.