### Quick Start Development Setup

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/docs/TDD.md

Basic setup instructions for starting the development server. This includes navigating to the site directory, installing dependencies, setting up the .env file, and running the development server.

```bash
cd site
npm install
cp .env.example .env # Add your OPENAI_API_KEY

# Development server (uses cached content)

npm run dev

# Full build with AI generation

npm run build
```

--------------------------------

### Install Dependencies and Run Development Server

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/CONTRIBUTING.md

Install project dependencies, copy the environment example, and start the development server. Ensure OPENAI_API_KEY is set in the .env file for AI generation.

```bash
pnpm install
cp .env.example .env  # Add OPENAI_API_KEY for AI generation
pnpm run dev
```

--------------------------------

### Install and Run Template Site Dev Server

Source: https://github.com/comfy-org/workflow_templates/blob/main/README.md

Navigate to the site directory, install dependencies using pnpm, and start the development server.

```bash
cd site
pnpm install
pnpm run dev       # Start dev server at localhost:4321
```

--------------------------------

### Install Dependencies and Run Development Server

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/README.md

Installs project dependencies, copies the environment file, and starts the development server. Ensure you have Node.js 18+ and pnpm 9.15+ installed.

```bash
pnpm install

cp .env.example .env

pnpm run dev
```

--------------------------------

### Test Installation from TestPyPI

Source: https://github.com/comfy-org/workflow_templates/blob/main/docs/publishing-workflow-templates.md

Create a virtual environment, install packages from TestPyPI, and verify the installation by checking the number of available templates. This ensures packages are correctly published and accessible.

```bash
python -m venv /tmp/testenv && source /tmp/testenv/bin/activate
python -m pip install --upgrade pip
python -m pip install --index-url https://test.pypi.org/simple --extra-index-url https://pypi.org/simple comfyui-workflow-templates
```

```python
import comfyui_workflow_templates_core; print(len(list(comfyui_workflow_templates_core.iter_templates())))
```

--------------------------------

### Install Dependencies

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Installs all project dependencies. This command is required before running any other commands.

```bash
pnpm install
```

--------------------------------

### Example Report Output

Source: https://github.com/comfy-org/workflow_templates/blob/main/scripts/README_whitelist.md

This is an example of the report generated by the script, showing whitelist configuration details and summary statistics of the node check.

```text
# ComfyUI Template Third-Party Node Check Report

## Whitelist Configuration
- Allowed cnr_ids: comfy-core, your-custom-node
- Allowed node_types: KSampler, CheckpointLoaderSimple
- Custom whitelisted nodes: 1

## Summary
- Total files checked: 168
- Files with third-party nodes: 0
- Total third-party nodes: 0
- Total whitelisted nodes: 1546

## ✅ All templates use official or whitelisted nodes
```

--------------------------------

### Example Template File Naming

Source: https://github.com/comfy-org/workflow_templates/blob/main/README.md

An example illustrating the naming convention for a 'text to video' template using Wan 2.1, including the JSON file and its corresponding WebP thumbnail.

```plaintext
text_to_video_wan.json
text_to_video_wan-1.webp
```

--------------------------------

### Install build and twine with pipx

Source: https://github.com/comfy-org/workflow_templates/blob/main/docs/publishing-workflow-templates.md

Installs the 'build' and 'twine' Python packages using pipx, a tool for installing and running Python applications in isolated environments.

```bash
pipx install build
pipx install twine
```

--------------------------------

### Start Development Server

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Starts the Astro development server, which typically runs on localhost:4321. Useful for local development and testing.

```bash
pnpm run dev
```

--------------------------------

### Install build and twine with pip

Source: https://github.com/comfy-org/workflow_templates/blob/main/docs/publishing-workflow-templates.md

Installs or upgrades the 'build' and 'twine' Python packages using pip within a virtual environment or globally.

```bash
python -m pip install --upgrade build twine
```

--------------------------------

### Install ComfyUI Subgraph Blueprints

Source: https://github.com/comfy-org/workflow_templates/blob/main/packages/blueprints/README.md

Install the library using pip. This command is used to add the package to your Python environment.

```bash
pip install comfyui-subgraph-blueprints
```

--------------------------------

### Input File Naming Convention Examples

Source: https://github.com/comfy-org/workflow_templates/blob/main/scripts/check_input_assets.md

Examples of input file naming conventions. The script uses this pattern to extract workflow names and descriptions, generating display names and tags for the upload JSON.

```plaintext
image_to_video_input_image.png
flux_dev_checkpoint_example_prompt.txt
audio_stable_audio_example_reference.mp3
```

--------------------------------

### Environment Variables Example

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/docs/TDD.md

Example of environment variables used in the .env file. OPENAI_API_KEY is required for AI content generation, and SKIP_AI_GENERATION can be used to skip AI calls.

```bash
# site/.env.example
OPENAI_API_KEY=sk-...  # Required for AI content generation

# Optional (for full builds)
SKIP_AI_GENERATION=true  # Skip AI calls, use cached/placeholder content
```

--------------------------------

### Vidu3ImageToVideoNode

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/knowledge/nodes/api/Vidu3ImageToVideoNode.md

Generates a video sequence starting from an input image using the Vidu Q3 Pro model. The video can be optionally guided by a text prompt and includes settings for resolution, duration, and audio.

```APIDOC
## Vidu3ImageToVideoNode

### Description

Generates a video sequence starting from an input image using the Vidu Q3 Pro model. The video can be optionally guided by a text prompt and includes settings for resolution, duration, and audio.

### Inputs

- **model** (COMBO) - Required - Model to use for video generation. Allowed values: `"viduq3-pro"`.
- **model.resolution** (COMBO) - Required - Resolution of the output video. Allowed values: `"720p"`, `"1080p"`, `"2K"`.
- **model.duration** (INT) - Required - Duration of the output video in seconds. Range: 1 to 16. Default: 5.
- **model.audio** (BOOLEAN) - Required - When enabled, outputs video with sound. Allowed values: `True`, `False`. Default: `False`.
- **image** (IMAGE) - Required - An image to be used as the start frame of the generated video. The image must have an aspect ratio between 1:4 and 4:1.
- **prompt** (STRING) - Optional - An optional text prompt for video generation. Maximum 2000 characters. Default: empty.
- **seed** (INT) - Optional - A seed value for controlling the randomness of the generation. Range: 0 to 2147483647. Default: 1.

### Outputs

- **output** (VIDEO) - The generated video file.
```

--------------------------------

### Sync Tutorials Command

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/docs/ROADMAP.md

Use this command to synchronize tutorial content from the documentation repository.

```bash
# Sync tutorials from docs repo
pnpm run sync:tutorials
```

--------------------------------

### Vidu2StartEndToVideoNode

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/knowledge/nodes/api/Vidu2StartEndToVideoNode.md

Generates a video by interpolating between a provided start frame and an end frame, guided by a text prompt. It uses a specified Vidu model to create a smooth transition between the two images over a set duration.

```APIDOC
## Vidu2StartEndToVideoNode

### Description
Generates a video by interpolating between a provided start frame and an end frame, guided by a text prompt. It uses a specified Vidu model to create a smooth transition between the two images over a set duration.

### Inputs

- **model** (COMBO) - Required - The Vidu model to use for video generation. Possible values: "viduq2-pro-fast", "viduq2-pro", "viduq2-turbo".
- **first_frame** (IMAGE) - Required - The starting image for the video sequence. Only a single image is allowed.
- **end_frame** (IMAGE) - Required - The ending image for the video sequence. Only a single image is allowed.
- **prompt** (STRING) - Required - A text description guiding the video generation (maximum 2000 characters).
- **duration** (INT) - Optional - The length of the generated video in seconds (default: 5). Range: 2 to 8.
- **seed** (INT) - Optional - A number used to initialize the random generation for reproducible results (default: 1). Range: 0 to 2147483647.
- **resolution** (COMBO) - Optional - The output resolution of the generated video. Possible values: "720p", "1080p".
- **movement_amplitude** (COMBO) - Optional - The movement amplitude of objects in the frame. Possible values: "auto", "small", "medium", "large".

**Note:** The `first_frame` and `end_frame` images must have similar aspect ratios. The node will validate that their aspect ratios are within a relative range of 0.8 to 1.25.

### Outputs

- **output** (VIDEO) - The generated video file.
```

--------------------------------

### Sync Tutorials

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Synchronizes tutorial content from the docs repository. This builds the knowledge base used for AI context.

```bash
pnpm run sync:tutorials
```

--------------------------------

### Full Site Build

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Executes a full build process, including syncing data, generating AI content, and creating previews. This is the command used for production deployment.

```bash
pnpm run build
```

--------------------------------

### Wan 2.1 Text-to-Video Workflow Example

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/knowledge/prompts/breakthrough.md

This JSON object details the breakthrough content for the Wan 2.1 text-to-video workflow. It includes an extended description of the feature, instructions on how to use it, a meta description for search engines, suggested use cases, and frequently asked questions.

```json
{
  "extendedDescription": "Wan 2.1 introduces locally-run text-to-video generation in ComfyUI, released in early 2025. This workflow converts a text prompt into a 480p video clip of up to 5 seconds, running entirely on consumer hardware without cloud API calls.\n\nPreviously, text-to-video required cloud services or research-grade GPUs. Wan 2.1's 14B parameter model brings comparable output quality to a single desktop GPU with 12 GB VRAM. The model produces temporally consistent motion with fewer flickering artifacts than earlier open-source video models.\n\nFor creators who need quick video drafts, motion tests, or animated concept pieces, this workflow eliminates the cost and latency of cloud generation. Results are best at 480p; higher resolutions may be possible in future model updates.",
  "howToUse": [
    "Ensure you have ComfyUI version 0.3.10 or later installed",
    "Download wan2.1_t2v_14B_fp16.safetensors and place it in models/diffusion_models",
    "Download umt5_xxl_fp8_e4m3fn_scaled.safetensors and place it in models/text_encoders",
    "Enter your scene description in the CLIP Text Encoder node",
    "Set frame count in the EmptyHunyuanLatentVideo node (default 49 frames for ~2 seconds)",
    "Click Queue or press Ctrl+Enter to generate the video"
  ],
  "metaDescription": "Wan 2.1 text-to-video now in ComfyUI. Generate 480p video clips from text prompts on consumer GPUs. Local, private, no cloud required.",
  "suggestedUseCases": [
    "Generate motion tests for animation pre-production",
    "Create short video loops for social media posts",
    "Produce animated concept art without manual keyframing",
    "Draft video ad storyboards from text descriptions"
  ],
  "faqItems": [
    {
      "question": "What's new in Wan 2.1 compared to previous video models?",
      "answer": "Wan 2.1 offers improved temporal consistency and reduced flickering compared to earlier open-source video models. The 14B parameter architecture produces more natural motion, and the model runs on consumer GPUs with 12 GB VRAM rather than requiring data-center hardware."
    },
    {
      "question": "Do I need to update ComfyUI for Wan 2.1?",
      "answer": "Yes, Wan 2.1 requires ComfyUI version 0.3.10 or later. The model uses custom node types not present in earlier versions. Update your installation before downloading the model weights."
    }
  ]
}
```

--------------------------------

### Build all wheels and run validation

Source: https://github.com/comfy-org/workflow_templates/blob/main/docs/publishing-workflow-templates.md

Checks out the main branch, pulls the latest changes, regenerates manifests, builds all wheels into the ./dist/ directory, and runs linting, tests, and 'twine check'.

```bash
git checkout main
git pull

./run_full_validation.sh
```

--------------------------------

### SEO Keyword Strategy - H1 and H2 Examples

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/docs/TDD.md

Provides examples of how H1 and H2 tags should be structured for SEO, incorporating template titles and keywords.

```html
<h1>Isometric Miniatures from a Selfie – ComfyUI Workflow</h1>
<h2>How to Generate Isometric Figurines with Nano Banana Pro</h2>
<h2>Step-by-Step: Upload Your Photo and Create Character Turnarounds</h2>
<h2>What You'll Need</h2>
<h2>Frequently Asked Questions</h2>
```

--------------------------------

### Set Up Pre-commit Hooks

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/CONTRIBUTING.md

Prepare the project to automatically format staged files on commit using Husky and lint-staged. This command sets up the Git hooks.

```bash
pnpm run prepare  # Sets up Git hooks (runs automatically on install)
```

--------------------------------

### Preview Production Build

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Locally previews the production build of the site. This helps in testing the final output before deployment.

```bash
pnpm run preview
```

--------------------------------

### Workflow JSON Model Metadata Example

Source: https://github.com/comfy-org/workflow_templates/blob/main/docs/SPEC.md

Example of how to embed model metadata within a workflow JSON file. This format is used for automatic model downloads.

```json
{
  "properties": {
    "models": [
      {
        "name": "model_filename.safetensors",
        "url": "https://huggingface.co/...",
        "hash": "sha256_hash",
        "hash_type": "SHA256",
        "directory": "models/checkpoints"
      }
    ]
  }
}
```

--------------------------------

### Basic Script Execution

Source: https://github.com/comfy-org/workflow_templates/blob/main/scripts/README_whitelist.md

Run the script to check templates using default configurations or specify custom directories and whitelist files.

```bash
# Check templates with default configuration
python3 scripts/check_third_party_nodes.py

# Specify template directory
python3 scripts/check_third_party_nodes.py --templates-dir ./my_templates

# Specify whitelist configuration file
python3 scripts/check_third_party_nodes.py --whitelist ./my_whitelist.json
```

--------------------------------

### Build Template Site for Production

Source: https://github.com/comfy-org/workflow_templates/blob/main/README.md

Build the Astro static site for production. This command automatically runs the prebuild pipeline.

```bash
pnpm run build     # Production build (runs prebuild pipeline automatically)
```

--------------------------------

### Full Prebuild Pipeline Execution

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/AGENTS.md

Executes the prebuild pipeline, which includes syncing templates and tutorials, and generating AI content and previews. This runs automatically before the main build.

```bash
pnpm run build:profile
```

```bash
pnpm run prebuild:sequential
```

--------------------------------

### KlingImageToVideoWithAudio Node

Source: https://github.com/comfy-org/workflow_templates/blob/main/site/knowledge/nodes/api/KlingImageToVideoWithAudio.md

Generates a video from a starting image and a text prompt using the Kling AI model. It can optionally include AI-generated audio.

```APIDOC
## KlingImageToVideoWithAudio Node

### Description
Generates a short video from a single starting image and a text prompt using the Kling AI model. The video begins with the provided image and can optionally include AI-generated audio.

### Inputs

#### Input Parameters

- **`model_name`** (COMBO) - Required - The specific version of the Kling AI model to use for video generation. Allowed values: `"kling-v2-6"`.
- **`start_frame`** (IMAGE) - Required - The image that will serve as the first frame of the generated video. The image must be at least 300x300 pixels and have an aspect ratio between 1:2.5 and 2.5:1.
- **`prompt`** (STRING) - Required - Positive text prompt describing the video content. The prompt must be between 1 and 2500 characters long.
- **`mode`** (COMBO) - Required - The operational mode for the video generation. Allowed values: `"pro"`.
- **`duration`** (COMBO) - Required - The length of the video to generate, in seconds. Allowed values: `5`, `10`.
- **`generate_audio`** (BOOLEAN) - Optional - When enabled, the node will generate audio to accompany the video. When disabled, the video will be silent. (default: True)

### Outputs

#### Output Parameters

- **`video`** (VIDEO) - The generated video file, which may include audio depending on the `generate_audio` input.
```