ComfyDownloader

简体中文

A desktop application for downloading AI models to the correct ComfyUI directories, powered by aria2.

Why This Project

ComfyDownloader is built for the annoying part of the ComfyUI workflow: finding a model, figuring out what type it is, choosing the correct subdirectory, and downloading it reliably.

Instead of juggling browser tabs, model pages, and manual folder selection, the app focuses on a tighter flow:

• Search models from Hugging Face and Civitai in one desktop app
• Parse URLs and suggest the correct ComfyUI model type
• Route downloads into the right model directory
• Track resumable downloads with aria2
• Inspect workflow files to see which models are missing locally

Screenshots

Search

Downloads

Workflow

Settings

Features

• Smart URL parsing - Automatically extracts filenames and metadata from Civitai, Hugging Face, and direct links
• Model type recommendation - Suggests the correct model type based on filename, URL, and source API metadata
• Directory mapping - Automatically routes downloads to the correct ComfyUI model subdirectory
• aria2-powered downloads - Concurrent, resumable downloads with real-time progress tracking
• Duplicate detection - Checks for existing files before downloading
• Batch import - Import multiple URLs at once from text or file
• Custom rules - Define keyword-based rules for automatic model type classification
• Proxy support - HTTP/SOCKS5 proxy for downloading behind firewalls
• ComfyUI integration - Verify models are recognized by a running ComfyUI instance
• Workflow parser - Analyze ComfyUI workflow JSON and check which models are missing locally

Quick Start

1. Download the latest build from Releases.
2. Open the app and configure your ComfyUI root or model base directory.
3. Paste a model URL or search from Hugging Face / Civitai.
4. Confirm the suggested model type and target directory.
5. Start the download and track progress from the Downloads page.

Platform Support

Pre-built binaries are currently published for:

• macOS (Apple Silicon, aarch64) - .dmg
• macOS (Intel, x86_64) - .dmg
• Windows (x86_64) - .msi / .exe

Linux packaging is planned, but it is not officially supported yet.

Each release also ships a SHA256SUMS.txt file for artifact verification.

These builds are not code-signed yet. See First-time launch on macOS / Windows.

Verifying the download (SHA256)

# macOS / Linux
shasum -a 256 ComfyDownloader_*.dmg

# Windows (PowerShell)
Get-FileHash .\ComfyDownloader_*.msi -Algorithm SHA256

Compare the output against the matching line in SHA256SUMS.txt.

First-time launch on macOS / Windows

Because the project does not yet use Apple Developer ID signing or a Windows EV certificate, the OS may warn you on first launch. The binaries are built from this repository by GitHub Actions, and you can inspect the workflow in release.yml.

macOS - If you see "ComfyDownloader is damaged and can't be opened" or "Apple cannot check it for malicious software", this is usually Gatekeeper blocking an unsigned app, not a corrupted DMG.

1. Drag ComfyDownloader.app into Applications
2. Run:

xattr -dr com.apple.quarantine /Applications/ComfyDownloader.app

3. Open the app again

You can also right-click the app and choose Open.

Windows - If SmartScreen shows "Windows protected your PC", click More info -> Run anyway.

Tech Stack

• Frontend: React 19 + TypeScript + Tailwind CSS v4 + shadcn/ui
• Backend: Tauri v2 (Rust)
• Download Engine: aria2c (embedded sidecar)
• Database: SQLite via tauri-plugin-sql
• State Management: Zustand

Development

Prerequisites

• Node.js >= 20
• Rust >= 1.77
• Windows 10+ or macOS

Local setup

# Install dependencies
npm install

# Re-download aria2c if needed
npm run prepare:aria2

# Start Vite + Tauri
npm run tauri dev

# Build production app
npm run tauri build

npm install automatically downloads the required aria2c sidecar into src-tauri/binaries/ for supported Windows and macOS targets.

Useful checks:

npm run lint
npm run typecheck
cd src-tauri && cargo test

Security Notes

• Hugging Face and Civitai access tokens are currently stored in the local SQLite settings database. Avoid reusing highly privileged tokens.
• Releases are not code-signed yet. Verify checksums before first launch.
• For security-sensitive reports, follow SECURITY.md instead of opening a public issue.

License

ComfyDownloader is released under the MIT License.

The project bundles aria2c as a sidecar download engine. aria2c is distributed under GNU GPL v2.0 or later and remains the property of its upstream authors. See THIRD_PARTY_NOTICES.md for the full list of third-party components and their licenses.

Project Structure

src/                    # React frontend
  components/           # UI components
  pages/                # Downloads, Search, Workflow, Settings
  stores/               # Zustand state management
  hooks/                # Custom React hooks
  lib/                  # Types, API wrappers, utilities

src-tauri/              # Rust backend
  src/
    aria2/              # aria2 process management and RPC client
    commands/           # Tauri command handlers
    url_parser/         # URL parsing for Civitai, Hugging Face, etc.
    model_type/         # Model type recommendation engine
    db/                 # SQLite migrations
  binaries/             # aria2c sidecar binary

Roadmap

• Improve release polish for public users
• Expand workflow parsing coverage for more ComfyUI node patterns
• Improve token handling and local secret storage
• Add official Linux packaging support
• Add richer screenshots and onboarding docs

Contributing

Contributions are welcome, including bug fixes, workflow parser improvements, and i18n updates.

• Start from the dev branch for normal contribution work
• Conventional Commits are preferred (feat:, fix:, docs:, chore:)
• Opening an issue first is optional for straightforward fixes

See CONTRIBUTING.md, SECURITY.md, and CODE_OF_CONDUCT.md for more details.