A simple wrapper that facilitates using ComfyUI as a stateless API, either by receiving images in the response, or by sending completed images to a webhook

• ComfyUI API - A Stateless and Extendable API for ComfyUI
  • Download and Use
  • Features
  • Full ComfyUI Support
  • Stateless API
    • Request Format
    • Response Format
    • Example Usage
      • Base64 Response
      • Webhook Response with Base64 Images
      • S3 Urls in Response
      • S3 Urls in Webhook Payload
      • Azure Blob Urls in Response
      • Azure Blob Urls in Webhook Payload
  • Model Manifest
  • Downloading Behavior
  • LRU Caching
  • Modular Storage Backends
    • S3-Compatible Storage
    • Huggingface Repository
    • Azure Blob Storage
    • HTTP
  • Image To Image Workflows
  • Dynamic Model Loading
  • Server-side image processing
  • Probes
  • API Configuration Guide
    • Environment Variables
    • Configuration Details
    • Additional Notes
  • Using Synchronously
  • Using with Webhooks
    • prompt.complete
    • prompt.failed
    • Validating Webhooks
      • Node.js Example
      • Python Example
    • DEPRECATED: Legacy Webhook Behavior
      • output.complete
      • prompt.failed (legacy)
  • System Events
    • status
    • progress
    • executing
    • execution_start
    • execution_cached
    • executed
    • execution_success
    • execution_interrupted
    • execution_error
    • file_downloaded
    • file_uploaded
    • file_deleted
  • Prebuilt Docker Images
  • Considerations for Running on SaladCloud
  • Custom Workflows
  • Contributing
  • Architecture

Either use a pre-built Docker image, or build your own.

Download the latest version from the release page, and copy it into your existing ComfyUI dockerfile. You can find good base dockerfiles in the docker directory. There are also example dockerfiles for popular models in the SaladCloud Recipes Repo.

If you have your own ComfyUI dockerfile, you can add the comfyui-api server to it like so:

# Change this to the version you want to use
ARG api_version=1.13.3

# Download the comfyui-api binary, and make it executable
ADD https://github.com/SaladTechnologies/comfyui-api/releases/download/${api_version}/comfyui-api .
RUN chmod +x comfyui-api

# Set CMD to launch the comfyui-api binary. The comfyui-api binary will launch ComfyUI as a child process.
CMD ["./comfyui-api"]

The server will be available on port 3000 by default, but this can be customized with the PORT environment variable.

The server hosts swagger docs at /docs, which can be used to interact with the API.

• Full Power Of ComfyUI: The server supports the full ComfyUI /prompt API, and can be used to execute any ComfyUI workflow.
• Verified Model/Workflow Support: Stable Diffusion 1.5, Stable Diffusion XL, Stable Diffusion 3.5, Flux, AnimateDiff, LTX Video, Hunyuan Video, CogVideoX, Mochi Video, Cosmos 1.0. My assumption is more model types are supported, but these are the ones I have verified.
• Stateless API: The server is stateless, and can be scaled horizontally to handle more requests.
• Swagger Docs: The server hosts swagger docs at /docs, which can be used to interact with the API.
• "Synchronous" Support: The server will return base64-encoded images directly in the response, if no async behavior is requested.
• Async Support via Webhooks: The server can send completed outputs to a webhook URL, allowing for asynchronous processing.
• Modular Storage Backends: Completed outputs can be sent base64-encoded to a webhook, or uploaded to any s3-compatible storage, an http endpoint, a huggingface repo, or azure blob storage. All of these can be used to download input media as well. More storage backends can be added easily. Supports an optional LRU cache for downloaded models and files to keep local storage from overflowing.
• Warmup Workflow: The server can be configured to run a warmup workflow on startup, which can be used to load and warm up models, and to ensure the server is ready to accept requests.
• Return Images In PNG (default), JPEG, or WebP: The server can return images in PNG, JPEG, or WebP format, via a parameter in the API request. Most options supported by sharp are supported.
• Probes: The server has two probes, /health and /ready, which can be used to check the server's health and readiness to receive traffic.
• Dynamic Workflow Endpoints: Automatically mount new workflow endpoints by adding conforming .js or .ts files to the /workflows directory in your docker image. See the guide for more information. A Claude 4 Sonnet prompt is included to assist in automating this process.
• Bring Your Own Models And Extensions: Use any model or extension you want by adding them to the normal ComfyUI directories /opt/ComfyUI/. You can configure a manifest file to download models and install extensions automatically on startup.
• Dynamic Model Loading: If you provide a URL in a model-loading node, the server will locally cache the model automatically before executing the workflow.
• Execution Stats: The server will return execution stats in the response.
• Works Great with SaladCloud: The server is designed to work well with SaladCloud, and can be used to host ComfyUI on the SaladCloud platform. It is likely to work well with other platforms as well.
  • Manages Deletion Cost: ONLY ON SALAD. The server will automatically set the instance deletion cost to the queue length, so that busier nodes are less likely to be scaled in while they are processing requests.
• Single Binary: The server is distributed as a single binary, and can be run with no dependencies.
• Websocket Events Via Webhook: The server can forward ComfyUI websocket events to a configured webhook, which can be used to monitor the progress of a workflow.
• Friendly License: The server is distributed under the MIT license, and can be used for any purpose. All of its dependencies are also MIT or Apache 2.0 licensed, except ComfyUI itself, which is GPL-3.0 licensed.

ComfyUI API sits in front of ComfyUI, and uses the ComfyUI /prompt API to execute workflows, so any API-formatted prompt can be executed by the server. Before queueing the prompt, the server will download any required inputs, such as images. It also overrides the filename_prefix field in the prompt to ensure that output files are saved with a unique filename. Once the prompt is queued, the server will wait for the prompt to complete, and then return the outputs in the response body, via a webhook, or upload them to S3, depending on the request parameters. Because of this, anything you can run in ComfyUI can be run in the ComfyUI API server, including custom nodes and workflows, and any models ComfyUI supports.

The ComfyUI API server is designed to be stateless, meaning that it does not store any state between requests. This allows the server to be scaled horizontally behind a load balancer, and to handle more requests by adding more instances of the server. The server uses a configurable warmup workflow to ensure that ComfyUI is ready to accept requests, and to load any required models. The server also self-hosts swagger docs and an openapi spec at /docs, which can be used to interact with the API.

Prompts are submitted to the server via the POST /prompt endpoint, which accepts a JSON body containing the prompt graph, as well as any additional parameters such as the webhook URL, S3 bucket and prefix, and image conversion options. A request may look something like:

{
  "id": "123e4567-e89b-12d3-a456-426614174000",
  "prompt": {
    "1": {
      "inputs": {
        "image": "https://salad-benchmark-assets.download/coco2017/train2017/000000000009.jpg",
        "upload": "image"
      },
      "class_type": "LoadImage"
    }
  },
  "webhook_v2": "https://example.com/webhook",
  "convert_output": {
    "format": "jpeg",
    "options": {
      "quality": 80,
      "progressive": true
    }
  }
}

• Only the prompt field is required. The other fields are optional, and can be omitted if not needed.
• Your prompt must be a valid ComfyUI prompt graph, which is a JSON object where each key is a node ID, and the value is an object containing the node's inputs, class type, and optional metadata.
• Your prompt must include a node that saves an output, such as a SaveImage node.

For async requests (i.e. when a webhook or S3 upload is used), the server will return a 202 Accepted response immediately, and the outputs will be sent to the webhook or uploaded to S3 in the background.

For synchronous requests (i.e. no webhook or s3.async is false), the server will return a 200 OK response once the prompt has completed, with a body containing the outputs. The response body will have the following format:

{
  "id": "123e4567-e89b-12d3-a456-426614174000",
  "prompt": { ... },
  "images": [
    "base64-encoded-image-1",
    "base64-encoded-image-2"
  ],
  "filenames": [
    "output-filename-1.png",
    "output-filename-2.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1625247600000,
        "end": 1625247605000,
        "duration": 5000
      },
      "nodes": {
        "1": {
          "start": 1625247600000
        },
        "2": {
          "start": 1625247601000
        }
      }
    },
    "preprocess_time": 1500,
    "upload_time": 1,
    "total_time": 6576
  }
}

If you requested image conversion, the images will be in the requested format (e.g. JPEG or WebP) instead of PNG.

Request:

{
  "prompt": { ... }
}

Response:

{
  "id": "generated-uuid",
  "prompt": { ... },
  "images": ["base64-encoded-image-1", "base64-encoded-image-2"],
  "filenames": ["generated-uuid_ComfyUI_0.png", "generated-uuid_ComfyUI_1.png"],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 1,
    "total_time": 6205
  }
}

Request:

{
  "prompt": { ... },
  "webhook_v2": "https://example.com/webhook"
}

HTTP Response: 202 Accepted

{
  "id": "generated-uuid",
  "status": "ok",
  "webhook_v2": "https://example.com/webhook",
  "prompt": { ... }
}

Webhook Payload

{
  "type": "prompt.complete",
  "id": "generated-uuid",
  "prompt": { ... },
  "webhook_v2": "https://example.com/webhook",
  "images": [
    "base64-encoded-image-1",
    "base64-encoded-image-2"
  ],
  "filenames": [
    "output-filename-1.png",
    "output-filename-2.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 1,
    "total_time": 6205
  }
}

Request:

{
  "prompt": { ... },
  "s3": {
    "bucket": "my-bucket",
    "prefix": "outputs/",
    "async": false
  }
}

Response:

{
  "id": "generated-uuid",
  "prompt": { ... },
  "images": [
    "s3://my-bucket/outputs/generated-uuid_ComfyUI_0.png",
    "s3://my-bucket/outputs/generated-uuid_ComfyUI_1.png"
  ],
  "filenames": [
    "generated-uuid_ComfyUI_0.png",
    "generated-uuid_ComfyUI_1.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 300,
    "total_time": 6505
  }
}

Request:

{
  "prompt": { ... },
  "s3": {
    "bucket": "my-bucket",
    "prefix": "outputs/"
  },
  "webhook_v2": "https://example.com/webhook"
}

HTTP Response: 202 Accepted

{
  "id": "generated-uuid",
  "status": "ok",
  "webhook_v2": "https://example.com/webhook",
  "s3": {
    "bucket": "my-bucket",
    "prefix": "outputs/",
  },
  "prompt": { ... }
}

Webhook Payload

{
  "type": "prompt.complete",
  "id": "generated-uuid",
  "prompt": { ... },
  "webhook_v2": "https://example.com/webhook",
  "s3": {
    "bucket": "my-bucket",
    "prefix": "outputs/"
  },
  "images": [
    "s3://my-bucket/outputs/generated-uuid_ComfyUI_0.png",
    "s3://my-bucket/outputs/generated-uuid_ComfyUI_1.png"
  ],
  "filenames": [
    "generated-uuid_ComfyUI_0.png",
    "generated-uuid_ComfyUI_1.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 300,
    "total_time": 6505
  }
}

Request:

{
  "prompt": { ... },
  "azure_blob_upload": {
    "container": "my-container",
    "blob_prefix": "outputs/",
    "async": false
  }
}

Response:

{
  "id": "generated-uuid",
  "prompt": { ... },
  "images": [
    "https://<your-account>.blob.core.windows.net/my-container/outputs/generated-uuid_ComfyUI_0.png",
    "https://<your-account>.blob.core.windows.net/my-container/outputs/generated-uuid_ComfyUI_1.png"
  ],
  "filenames": [
    "generated-uuid_ComfyUI_0.png",
    "generated-uuid_ComfyUI_1.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 300,
    "total_time": 6505
  }
}

Request:

{
  "prompt": { ... },
  "azure_blob_upload": {
    "container": "my-container",
    "blob_prefix": "outputs/"
  },
  "webhook_v2": "https://example.com/webhook"
}

HTTP Response: 202 Accepted

{
  "id": "generated-uuid",
  "status": "ok",
  "webhook_v2": "https://example.com/webhook",
  "azure_blob_upload": {
    "container": "my-container",
    "blob_prefix": "outputs/",
  },
  "prompt": { ... }
}

Webhook Payload

{
  "type": "prompt.complete",
  "id": "generated-uuid",
  "prompt": { ... },
  "webhook_v2": "https://example.com/webhook",
  "azure_blob_upload": {
    "container": "my-container",
    "blob_prefix": "outputs/"
  },
  "images": [
    "https://<your-account>.blob.core.windows.net/my-container/outputs/generated-uuid_ComfyUI_0.png",
    "https://<your-account>.blob.core.windows.net/my-container/outputs/generated-uuid_ComfyUI_1.png"
  ],
  "filenames": [
    "generated-uuid_ComfyUI_0.png",
    "generated-uuid_ComfyUI_1.png"
  ],
  "stats": {
    "comfy_execution": {
      "total": {
        "start": 1700000000000,
        "end": 1700000005000,
        "duration": 5000
      },
      "nodes": {
        "3": {
          "start": 1700000000000
        },
        ...
      }
    },
    "preprocess_time": 1200,
    "upload_time": 300,
    "total_time": 6505
  }
}

The server can be configured to download models and install extensions automatically on startup, by providing a manifest file in either JSON or YAML format. The manifest filepath can be provided via the MANIFEST environment variable, or the full manifest as a JSON string via the MANIFEST_JSON environment variable. If both are provided, the MANIFEST_JSON variable will take precedence.

The manifest file should have the following format (all fields are optional):

apt:
  - package1
  - package2
pip:
  - package3
  - package4
custom_nodes:
  - node-name-from-comfy-registry
  - https://github.com/username/repo
models:
  before_start:
    - url: https://example.com/model.ckpt
      local_path: /opt/ComfyUI/models/checkpoints/model1.ckpt
    - url: s3://my-bucket/path/to/model.safetensors
      local_path: /opt/ComfyUI/models/checkpoints/model2.safetensors
  after_start:
    - url: https://example.com/another-model.ckpt
      local_path: /opt/ComfyUI/models/checkpoints/model3.ckpt

If a manifest is provided, the server will perform the following in order:

1. Install any apt packages listed in the apt field.
2. Install any pip packages listed in the pip field. Uses uv, otherwise falls back to pip.
3. Install any custom nodes listed in the custom_nodes field, using the comfy cli tool if available and a plain string is provided, or by cloning the provided git repository if a URL is provided. If cloned, requirements.txt will be installed if it exists, using uv if available, otherwise falling back to pip.
4. Download any models listed in the models.before_start field, and save them to the specified local_path.
5. Start background downloading any models listed in the models.after_start field, and save them to the specified local_path. These downloads will be started in the background and will not block the server from accepting requests. This is useful for preloading less frequently used models.

When downloading files, whether via the manifest, image-to-image workflows, or dynamic model loading, the server will first check if the file already exists at the specified path. It does this by hashing the provided URL and looking for a matching file in the cache directory ($HOME/.cache/comfyui-api by default). For example, the url https://civitai.com/api/download/models/128713?type=Model&format=SafeTensor&size=pruned&fp=fp16 will always be saved in the cache as Pk6VSKLStckZydwGhX0bM8TqaqHEW9yt.safetensors. If a matching file is found, it will be used instead of downloading the file again. This helps to reduce bandwidth usage and speed up request times.

If the url is an S3 URL, the server will use the AWS SDK to download the file. This allows the server to access private S3 buckets (or S3-compatible buckets), as long as the appropriate AWS credentials are provided via environment variables.

If the url is a huggingface URL, the server will use the hf cli tool to download the file. This allows you to take advantage of high-speed xet storage, as well as other optimizations provided by huggingface.

If the url is an azure blob storage URL, the server will use the Azure SDK to download the file.

If the url is a regular http(s) URL, the server will use fetch to stream the file to disk. If the url has a file extension, the server will use that extension when saving the file. Otherwise, it will attempt to determine the file extension from the Content-Disposition or Content-Type headers.

All downloaded files live in the configured cache directory with a name taken as the first 32 characters of the URL hash plus the file extension, and are symbolically linked to the specified local path.

If a download for a given URL is already in progress, any subsequent requests for the same URL will wait for the first download to complete, and then use the downloaded file.

The server uses an LRU cache to manage the cache directory, which is used to store downloaded models and other files. It is configured to be disabled by default, but you can set a size via the LRU_CACHE_SIZE_GB environment variable. When the cache size exceeds the configured size, the server will delete the least recently used files until the cache size is below the configured size. Note: Cache-size is determined after a download completes, so actual cache size can temporarily exceed the configured size while downloads are in progress.

The server supports multiple storage backends for downloading models and input media, and uploading completed outputs. All uploads take a prefix of some kind, not a full path or URL.

All uploads can be handled synchronously or asynchronously, depending on the async field in the upload block of the request body.

• If async is true or omitted, the server will return a 202 Accepted response immediately, and the upload will be handled in the background.
• If async is false, the server will wait for the upload to complete before returning a 200 OK response with the uploaded urls in the response body.

If an upload for a particular url is in progress, a subsequent upload to the same url will abort the first request and take over the upload. This is rooted in the assumption that you want the latest version of any particular output.

Includes AWS S3, Cloudflare R2, etc. Uses the AWS SDK. Requires appropriate AWS credentials to be provided via environment variables. Used for URLs starting with s3://.

For downloads, use the format s3://bucket-name/path/to/file. For uploads, include the s3 field in the request body, like:

{
  "prompt": {...}, 
  "s3": { 
    "bucket": "my-bucket", 
    "prefix": "optional/prefix", 
    "async": false 
  }
}

Uses the hf cli tool. Requires the HF_TOKEN environment variable to be set with a valid Huggingface token. Used for URLs starting with https://huggingface.co/. Works with both public and private repos, model and dataset repos, and large files stored with xet storage.

For downloads, use the format https://huggingface.co/username/repo/resolve/revision/path/to/file or https://huggingface.co/datasets/username/repo/resolve/revision/path/to/file.

For uploads, include the hf_upload field in the request body, like

{
  "prompt": {}, 
  "hf_upload": { 
    "repo": "username/repo", 
    "repo_type": "dataset", 
    "revision": "main", 
    "directory": "test-source-images", 
    "async": false 
  }
}

The repo_type field can be either model or dataset, and defaults to model.

Uses the Azure SDK. Requires appropriate Azure credentials to be provided via environment variables. Used for URLs matching https://<your-account>.blob.core.windows.net/.

For downloads, use the format https://<your-account>.blob.core.windows.net/container/path/to/file.

For uploads, include the azure_blob_upload field in the request body, like:

{
  "prompt": {}, 
  "azure_blob_upload": { 
    "container": "my-container", 
    "blob_prefix": "optional/prefix", 
    "async": false 
  }
}

Uses Fetch. Supports custom headers via the HTTP_AUTH_HEADER_NAME and HTTP_AUTH_HEADER_VALUE environment variables. Basic auth can be used via the URL, i.e. https://username:[email protected].

For downloads, use any valid http(s) URL that is not matched by the other storage backends.

For uploads, makes a PUT request to the specified URL with the image as the body. Matches any other URL not matched by the other storage backends.

The ComfyUI API server supports image-to-image workflows, allowing you to submit an image and receive a modified version of that image in response. This is useful for tasks such as image in-painting, style transfer, and other image manipulation tasks.

To use image-to-image workflows, you can submit an image as a base64-encoded string, or a URL. The server will automatically detect the input type and process the image accordingly, using an appropriate storage provider if necessary.

Here's an example of doing this in a LoadImage node:

{
  "inputs": {
    "image": "https://salad-benchmark-assets.download/coco2017/train2017/000000000009.jpg",
    "upload": "image"
  },
  "class_type": "LoadImage",
  "_meta": {
    "title": "Load Image"
  }
}

The ComfyUI API server supports dynamic model loading, allowing you to specify a model URL in a model-loading node, and the server will automatically download and cache the model before executing the workflow. This is useful for workflows that need to potentially use a different model for each request. An example may be head-shot generation, which would specify a LoRA per person. The LoRA may be generated on-the-fly by another service, and provided to the ComfyUI API server via a URL.

{
  "inputs": {
    "ckpt_name": "https://civitai.com/api/download/models/76750?type=Model&format=SafeTensor&size=pruned&fp=fp16"
  },
  "class_type": "CheckpointLoaderSimple",
  "_meta": {
    "title": "Load Checkpoint"
  }
},

The ComfyUI API server uses the sharp library to process images. This allows you to return the images in different, more compact formats, such as JPEG or WebP. This can be accomplished by including the convert_output object in the request body, which can contain the following fields:

{
  "format": "jpeg|webp",
  "options": {}
}

Omitting the convert_output object will default to PNG format, which is lossless and has the best quality, but is also the largest in size.

JPEG options:

• quality: The quality of the JPEG image, between 1 and 100. Default is 80.
• progressive: Use progressive (interlace) scanning. Default is false.
• chromaSubsampling: Set to 4:4:4 to prevent chroma subsampling otherwise defaults to 4:2:0 chroma subsampling.
• optimizeCoding: Optimize the Huffman coding tables. Default is true.
• mozjpeg: use mozjpeg defaults, equivalent to { trellisQuantisation: true, overshootDeringing: true, optimizeScans: true, quantisationTable: 3 }
• trellisQuantisation: Use trellis quantization. Default is false.
• overshootDeringing: Use overshoot deringing. Default is false.
• optimizeScans: Optimize the scan order. Default is false.
• quantisationTable: Set the quantization table to use, 1 - 8. Default is 0.

WebP options:

• quality: The quality of the WebP image, between 1 and 100. Default is 80.
• alphaQuality: The quality of the alpha channel, between 0 and 100. Default is 100.
• lossless: Use lossless compression. Default is false.
• nearLossless: Use near-lossless compression. Default is false.
• smartSubsample: Use smart subsampling. Default is false.
• preset: named preset for preprocessing/filtering, one of default, picture, photo, drawing, icon, or text. Default is default.
• effort: CPU effort level, between 0 (fastest) and 6 (slowest). Default is 4.

The server has two probes, /health and /ready.

• The /health probe will return a 200 status code once the warmup workflow has completed. It will stay healthy as long as the server is running, even if ComfyUI crashes.
• The /ready probe will also return a 200 status code once the warmup workflow has completed. It will return a 503 status code if ComfyUI is not running, such as in the case it has crashed, but is being automatically restarted. If you have set MAX_QUEUE_DEPTH to a non-zero value, it will return a 503 status code if ComfyUI's queue has reached the maximum depth.

The following table lists the available environment variables and their default values. For historical reasons, the default values mostly assume this will run on top of an ai-dock image, but we currently provide our own more minimal image here in this repo.

If you are using the s3 storage functionality, make sure to set all of the appropriate environment variables for your S3 bucket, such as AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and AWS_REGION. The server will automatically use these to upload images to S3.

If you are using the huggingface storage functionality, make sure to set the HF_TOKEN environment variable with a valid Huggingface token with appropriate permissions.

If you are using the azure blob storage functionality, make sure to set all of the appropriate environment variables for your Azure account, such as AZURE_STORAGE_CONNECTION_STRING.

1. ComfyUI Settings:

   • The application uses the CMD environment variable to specify the command for launching ComfyUI.
   • ComfyUI is accessed at http://${DIRECT_ADDRESS}:${COMFYUI_PORT_HOST}.

2. Wrapper Settings:

   • The wrapper API listens on HOST:PORT.
   • It can be accessed at http://localhost:${PORT}.
   • Use an IPv6 address for HOST when deploying on Salad. This is the default behavior.

3. Startup Checks:

   • The application performs startup checks at intervals specified by STARTUP_CHECK_INTERVAL_S.
   • It will attempt up to STARTUP_CHECK_MAX_TRIES before giving up.

4. Directories:

   • The application uses the COMFY_HOME environment variable to locate the ComfyUI installation.
   • Output files are stored in OUTPUT_DIR.
   • Input files are read from INPUT_DIR.
   • Model files are located in MODEL_DIR.
   • Workflow files are stored in WORKFLOW_DIR. See below for more information.

5. Warmup Prompt:

   • If WARMUP_PROMPT_FILE is set, the application will load and parse a warmup prompt from this file.
   • The checkpoint used in this prompt can be used as the default for workflow models.

6. Models:

   • The application scans the MODEL_DIR for subdirectories and creates configurations for each model type found.
   • Each model type will have its directory path, list of available models, and a Zod enum for validation.
   • The model names are exposed via the GET /models endpoint, and via the config object throughout the application.

7. ComfyUI Description:

   • The application retrieves available samplers and schedulers from ComfyUI itself at startup. It does not take custom nodes or extensions into account.
   • This information is used to create Zod enums for validation in workflows, but is otherwise not used by the application.

• The application uses Zod for runtime type checking and validation of configuration values.
• The configuration includes setup for both the wrapper application and ComfyUI itself.

Remember to set these environment variables according to your specific deployment needs before running the application.

The default behavior of the API is to return an array of base64-encoded outputs in the response body. All that is needed to do this is to omit webhook and upload fields from the request body.

ComfyUI API sends two types of webhooks: System Events, which are emitted by ComfyUI itself, and Workflow Events, which are emitted by the API server. See System Events for more information on System Events.

If a user includes .webhook_v2 field in a request to /prompt or any of the workflow endpoints, the server will send any completed outputs to the webhook URL provided in the request. It will also send a webhook if the request fails.

For successful requests including the .webhook_v2 field, a single webhook request will be sent once the entire workflow has completed, containing all outputs. Webhooks are sent as Standard Webhooks, and can be validated using the WEBHOOK_SECRET environment variable and any standard webhook validation library such as svix.

The webhook type name for a completed prompt is prompt.complete. The webhook will have the same schema as the synchronous response, with the addition of the type and timestamp fields:

{
  "type": "prompt.complete",
  "timestamp": "2025-01-01T00:00:00Z",
  "id": "request-id",
  "images": ["base64-encoded-image-1", "base64-encoded-image-2"],
  "filenames": ["output-filename-1.png", "output-filename-2.png"],
  "prompt": {},
  "stats":{}
}

Note that if you include upload fields in your request, the .images field will contain the uploaded URLs instead of base64-encoded images.

The webhook type name for a failed request is prompt.failed. The webhook will have the following schema:

{
  "type": "prompt.failed",
  "timestamp": "2025-01-01T00:00:00Z",
  "error": "error-message",
  "id": "request-id",
  "prompt": {}
}

npm install svix

const { Webhook } = require('svix')

//Express.js middleware
function validateWebhookSignature(req, res, next) {
  const webhook = new Webhook(secret)
  try {
    webhook.verify(req.body, req.headers)
    next()
  } catch (error) {
    console.error('Webhook verification failed:', error)
    return res.status(401).send('Invalid signature')
  }
}

pip install svix

from fastapi import FastAPI, Request, HTTPException
from svix import Webhook
from typing import Any, Dict

async def validate_webhook(request: Request) -> Dict[str, Any]:
    """
    FastAPI Dependency to validate webhook signatures
    """
    try:
        # Get the raw body
        body = await request.body()

        # Create webhook instance
        webhook = Webhook(webhook_secret)

        # Verify the webhook signature
        payload = webhook.verify(body, dict(request.headers))

        return payload
    except Exception as e:
        print(f"Webhook verification failed: {e}")
        raise HTTPException(status_code=401, detail="Invalid webhook signature")

LEGACY BEHAVIOR: For successful requests including the now-deprecated .webhook field, every output from the workflow will be sent as individual webhook requests. That means if your request generates 4 images, you will receive 4 webhook requests, each with a single image. These webhooks are not signed, so we recommend migrating to the new .webhook_v2 field as soon as possible.

The webhook event name for a completed output is output.complete. The webhook will have the following schema:

{
  "event": "output.complete",
  "image": "base64-encoded-image",
  "id": "request-id",
  "filename": "output-filename.png",
  "prompt": {}
}

The webhook event name for a failed request is prompt.failed. The webhook will have the following schema:

{
  "event": "prompt.failed",
  "error": "error-message",
  "id": "request-id",
  "prompt": {}
}

ComfyUI emits a number of events over websocket during the course of a workflow. These can be configured to be sent to a webhook using the SYSTEM_WEBHOOK_URL and SYSTEM_WEBHOOK_EVENTS environment variables. Additionally, any environment variable starting with SYSTEM_META_ will be sent as metadata with the event. From version 1.13.0, these are signed, and can be validated using the WEBHOOK_SECRET environment variable and any standard webhook validation library such as svix. See above for examples.

All webhooks have the same format, which is as follows:

{
  "event": "event_name",
  "data": {},
  "metadata": {}
}

When running on SaladCloud, .metadata will always include lowercase versions of the Default Environment Variables.

The following events are available:

• "status"
• "progress"
• "executing"
• "execution_start"
• "execution_cached"
• "executed"
• "execution_success"
• "execution_interrupted"
• "execution_error"
• "file_downloaded"
• "file_uploaded"
• "file_deleted"

The SYSTEM_WEBHOOK_EVENTS environment variable should be a comma-separated list of the events you want to send to the webhook. If not set, no events will be sent.

The event name received in the webhook will be comfy.${event_name}, i.e. comfy.progress, or storage.${event_name} for file events.

Example:

export SYSTEM_WEBHOOK_EVENTS="progress,execution_start,execution_success,execution_error"

This will cause the API to send the progress, execution_start, execution_success, and execution_error events to the webhook.

The SYSTEM_META_* environment variables can be used to add metadata to the webhook events. For example:

export SYSTEM_META_batch=abc
export SYSTEM_META_purpose=testing

Will add {"batch": "abc", "purpose": "testing"} to the .metadata field on system webhooks.

The following are the schemas for the event data that will be sent to the webhook. This will populate the .data field on the webhook.

{
  "type": "status",
  "data": {
    "status": {
      "exec_info": {
        "queue_remaining": 3
      }
    }
  },
  "sid": "abc123"
}

{
  "type": "progress",
  "data": {
    "value": 45,
    "max": 100,
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "node": "42"
  },
  "sid": "xyz789"
}

{
  "type": "executing",
  "data": {
    "node": "42",
    "display_node": "42",
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000"
  },
  "sid": "xyz789"
}

{
  "type": "execution_start",
  "data": {
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "timestamp": 1705505423000
  },
  "sid": "xyz789"
}

{
  "type": "execution_cached",
  "data": {
    "nodes": ["42", "7", "13"],
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "timestamp": 1705505423000
  },
  "sid": "xyz789"
}

{
  "type": "executed",
  "data": {
    "node": "42",
    "display_node": "42",
    "output": {},
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000"
  },
  "sid": "xyz789"
}

{
  "type": "execution_success",
  "data": {
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "timestamp": 1705505423000
  },
  "sid": "xyz789"
}

{
  "type": "execution_interrupted",
  "data": {
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "node_id": "42",
    "node_type": "KSampler",
    "executed": []
  },
  "sid": "xyz789"
}

{
  "type": "execution_error",
  "data": {
    "prompt_id": "123e4567-e89b-12d3-a456-426614174000",
    "node_id": "42",
    "node_type": "KSampler",
    "executed": [],
    "exception_message": "CUDA out of memory. Tried to allocate 2.20 GiB",
    "exception_type": "RuntimeError",
    "traceback": "Traceback (most recent call last):\n  File \"nodes.py\", line 245, in sample\n    samples = sampler.sample(model, noise, steps)",
    "current_inputs": {
      "seed": 42,
      "steps": 20,
      "cfg": 7.5,
      "sampler_name": "euler"
    },
    "current_outputs": []
  },
  "sid": "xyz789"
}

{
  // Where the file was downloaded from
  "url": "https://example.com/model.safetensors",

  // Local path where the file was saved
  "local_path": "/opt/ComfyUI/models/model.safetensors",

  // Size of the downloaded file in bytes
  "size": 123456789,

  // Duration of the download in seconds
  "duration": 2.34
}

{
  // Local path of the file that was uploaded
  "local_path": "/opt/ComfyUI/output/image.png",

  // URL where the file was uploaded to
  "url": "s3://my-bucket/images/image.png",

  // Size of the uploaded file in bytes
  "size": 123456,

  // Duration of the upload in seconds
  "duration": 0.56
}

{
  // URL of the file that was deleted. Note there are edge cases where this may be unknown, and the value will be "unknown".
  "url": "s3://my-bucket/models/old_model.safetensors",

  // Local path of the file that was deleted
  "local_path": "/opt/ComfyUI/models/old_model.safetensors",

  // Size of the deleted file in bytes
  "size": 987654321
}

You can find ready-to-go docker images under Packages in this repository.

The images are tagged with the comfyui-api version they are built with, and the comfyui version they are built for, along with their pytorch version and CUDA version. There are versions for both CUDA runtime and CUDA devel, so you can choose the one that best fits your needs.

The tag pattern is ghcr.io/saladtechnologies/comfyui-api:comfy<comfy-version>-api<api-version>-torch<pytorch-version>-cuda<cuda-version>-<runtime|devel> where:

• <comfy-version> is the version of ComfyUI used
• <api-version> is the version of the comfyui-api server
• <pytorch-version> is the version of PyTorch used
• <cuda-version> is the version of CUDA used
• <runtime|devel> is whether the image is built with the CUDA runtime or the CUDA devel image. The devel image is much larger, but includes the full CUDA toolkit, which is required for some custom nodes.

If the tag doesn't have api<api-version>, it does not include the api, and is just the ComfyUI base image.

Included in the API images are the following utilities:

• git
• curl
• wget
• unzip
• ComfyUI
• comfy cli

All of SaladCloud's image and video generation recipes are built on top of these images, so you can use them as a base for your own workflows. For examples of using this with custom models and nodes, check out the Salad Recipes repository on GitHub.

• SaladCloud's Container Gateway has a 100s timeout. It is possible to construct very long running ComfyUI workflows, such as for video generation, that would exceed this timeout. In this scenario, you will need to either use a webhook to receive the results, or integrate with SaladCloud's Job Queues to handle long-running workflows.
• SaladCloud's maximum container image size is 35GB(compressed). The base comfyui-api image is around 3.25GB(compressed), so any models and extensions must fit in the remaining space.

Custom workflows offer a simple and powerful way to create new endpoints for your specific use cases which abstract away the complexities of the underlying ComfyUI node-based prompt format. You can create workflows in either javascript or typescript, and they can be as simple or complex as you need them to be. Workflows are loaded at runtime, even when you use the pre-compiled binary releases or docker images, so you can easily add new workflows without needing to rebuild the image.

See the guide on generating new workflow endpoints for more information.

Contributions are welcome! See the Development guide for more information on how to develop, test, and contribute to this project. ComfyUI is a powerful tool with MANY options, and it's likely that not all of them are currently well supported by the comfyui-api server. Please open an issue with as much information as possible about the problem you're facing or the feature you need. If you have encountered a bug, please include the steps to reproduce it, and any relevant logs or error messages. If you are able, adding a failing test is the best way to ensure your issue is resolved quickly. Let's make productionizing ComfyUI as easy as possible!

The server is built with Fastify, a fast and low overhead web framework for Node.js. It sits in front of ComfyUI, and provides a RESTful API for interacting with ComfyUI.