ComfyUI-SendToEagle-w-Metadata

[English] [日本語]

• A custom node for ComfyUI
• Sends images with metadata (PNGInfo) obtained from the input values of each node to Eagle (image management software)
• You can also add arbitrary Key:Value pairs to the metadata
• You can customize the tags to be registered in Eagle
• You can save arbitrary strings and metadata in the Eagle notes field

Installation

cd <ComfyUI directory>/custom_nodes
git clone https://github.com/watarika/ComfyUI-SendToEagle-w-Metadata.git

List of Custom Nodes

• Send to Eagle With Metadata
• Send to Eagle With Metadata (Extended)
• Create Extra MetaData

Send to Eagle With Metadata

• Saves the images received as input to storage as images with metadata (PNGInfo), and then sends them to Eagle
• For details on the actual metadata added, please refer to Metadata Added to Images
• If the file format is png or webp, the workflow is also saved in the image
  Note: If it is webp and lossless, the workflow may not be saved (under investigation)

Default Behavior

The default behavior is as follows:
If you want to change these behaviors, use the Send to Eagle With Metadata (Extended) node.

• The file name is saved in the format yyyyMMdd-hhmmss_SSSSSS + extension
  • If the batch size is 2 or more, (Batch index) is appended to the end to ensure that images executed in batches have unique file names
• Sent to the default Eagle folder
• Metadata is also saved in the Eagle notes field
• The sampler name is converted to the notation used on Civitai
  • Example: If the sampler is dpmpp_2m and the scheduler is karras, it becomes DPM++ 2M Karras
• The model hash value is not output in the metadata
• Metadata is obtained from the first processed KSampler node

Configurable Settings

Item	Description
file_format	Specifies the format of the file to be saved. You can specify png, jpeg, or webp
lossless_webp	Specifies whether to compress the image when file_format is webp lossless: Saves the image without compression lossy: Compresses and saves the image
quality	Specifies the quality when compressing the image. Specify a value between 0 and 100. Note: Only valid for jpeg or webp (lossy).
save_only_no_send	If true, the image is saved to storage but not sent to Eagle
tag_pattern	Specifies what to add to Eagle tags. None: No tags are added Positive Prompt: The positive prompt is added as a tag Positive Prompt, Negative Prompt: The positive and negative prompts are added as tags Memo: The content of memo is added as a tag Custom: The string specified by custom_tag_pattern is added as a tag Note: When registering the negative prompt, "n:" is added to the beginning of the tag name to distinguish it from the positive prompt tag
custom_tag_pattern	Specifies the tag pattern to be used when tag_pattern is Custom, separated by commas. The following settings are available: Image metadata: Positive prompt, Negative prompt, Steps, Sampler, CFG scale, Seed, Clip skip, Size, Model, Model hash, VAE, VAE hash, Batch index, Batch size Memo content: Memo Additional metadata: The Key defined in extra_metadata Note: Items that do not match the above are added as tags as they are Note: Metadata for which a value cannot be obtained is added with a value of - (e.g., "Model hash: -")
memo	Specifies the text to be saved in the Eagle notes field. The specified text is appended with "Memo:" and saved at the end of the notes field
extra_metadata	Allows you to add your own metadata to the saved image. Can be omitted if not needed. Specifies the metadata to be added using the Create Extra MetaData node.
positive / negative	This is an option input for explicitly specifying prompts, rather than automatically retrieving them from the workflow. Use this input when prompts cannot be automatically retrieved or when you wish to specify prompts yourself.

Batch Support

The inputs memo, custom_tag_pattern, extra_metadata, positive, and negative support batch images.
If an input is a list, the corresponding element is applied to each image in the batch index. If the list is shorter than the batch size, empty strings are used. If a single string is passed, the same string is applied to all images.

Node Output

• filepath: Outputs the full file paths of saved images (in list format)
• filepath_count: Outputs the number of saved images
• image: Outputs the input image as received

Send to Eagle With Metadata (Extended)

• Saves the images received as input to storage as images with metadata (PNGInfo), and then sends them to Eagle
• You can customize the default behavior of Send to Eagle With Metadata
• Added feature: You can save the workflow's JSON file to storage

Additional Configurable Settings

• In addition to the settings that can be specified in Send to Eagle With Metadata, the following settings can be specified

Item	Description
filename_prefix	Specifies the string to be added to the beginning of the saved file name. You can also specify a directory by separating it with a slash (/). You can replace it with the acquired information by using %<Key>%. Note: Refer to the table below for the keys that can be specified.
sampler_selection_method	Specifies how to select the KSampler node executed before this node Farthest: Selects the KSampler node farthest from this node Nearest: Selects the KSampler node closest to this node By node ID: Selects the KSampler node with the node ID sampler_selection_node_id
sampler_selection_node_id	Only specified when sampler_selection_method is By node ID. Specifies the node ID of the KSampler node executed before this node. The node ID is the number displayed in the upper right corner of the node.
save_workflow_json	Specifies whether to save the workflow's JSON file to storage. Note: It will not be sent to Eagle.
add_counter_to_filename	Specifies whether to add a counter to the file name. true: Adds a sequential number to a group of files with the same file name excluding the counter part false: Does not add a counter. However, if the batch size is 2 or more, (Batch index) is appended to the end of the file name to ensure that images executed in batches have unique file names
civitai_sampler	Converts the sampler name to the notation used on Civitai Example: If the sampler is dpmpp_2m and the scheduler is karras, it becomes DPM++ 2M Karras
calc_model_hash	If true, calculates the hash value of the model and adds it to the metadata Note: It takes time to calculate the hash value of the model
send_metadata_as_memo	true: Saves the metadata in the Eagle notes field. If there is input to memo, its content is appended with "Memo:" and saved after the metadata false: Does not save the metadata in the Eagle notes field. Only the content of memo is saved in the Eagle notes field
eagle_folder	Specifies the image save destination folder in Eagle by FolderID or Folder name

Batch Support

The filename_prefix and eagle_folder inputs support batch images.
If an input is a list, elements corresponding to the batch index are applied. If the list is shorter than the batch size, empty strings are used. If a single string is passed, the same string is applied to all images.

Keys that can be specified in filename_prefix

Key	Replaced Information
%seed%	Seed value
%width%	Image width
%height%	Image height
%pprompt%	Positive Prompt
%pprompt:chars%	First n characters of Positive Prompt
%nprompt%	Negative Prompt
%nprompt:chars%	First n characters of Negative Prompt
%model%	Checkpoint name
%model:chars%	First n characters of Checkpoint name
%date%	Generation date and time (yyyyMMddhhmmss)
%date:%	Generation date and time

Refer to the table below for the identifiers specified in <format> of %date:<format>%.

Identifier	Description
yyyy	Year
MM	Month
dd	Day
hh	Hour
mm	Minute
ss	Second

Create Extra Metadata

• Specifies the metadata to be added to the saved image.

Metadata Added to Images

Item	Description
Positive prompt	Positive prompt
Negative prompt	Negative prompt
Steps	Number of steps
Sampler	Sampler + Scheduler name
CFG scale	CFG scale
Seed	Seed value
Clip skip	Clip skip value Note: Output only when the CLIP Set Last Layer node is used
Size	Image width × Image height
Model	Checkpoint name
Model hash	Checkpoint hash value Note: Output only when calc_model_hash is true
VAE	VAE name
VAE hash	VAE hash value Note: Output only when calc_model_hash is true
Lora_n Model name	nth LoRA name
Lora_n Model hash	nth LoRA hash value
Lora_n Strength model	Strength value for the model of the nth LoRA
Lora_n Strength clip	Strength value for the clip of the nth LoRA
Embedding_n Name	nth Embedding name
Embedding_n Hash	nth Embedding hash value
Batch index	Batch process index Note: Output only when Batch size >= 2
Batch size	Batch size Note: Output only when Batch size >= 2
Hashes	Outputs the hash values of Model, Lora_n, and Embedding_n separated by commas (for Civitai) Note: Output only when calc_model_hash is true
(Additional metadata)	Unique metadata entered in extra_metadata

Supported Nodes and Extensions

• Metadata is obtained from the inputs of the KSampler node found by sampler_selection_method and the inputs of previously executed nodes
  • The target KSampler nodes are the keys of SAMPLERS in py/defs/samplers.py and files under py/defs/ext/
• Please check the following files for supported nodes.
  • py/defs/captures.py
  • py/defs/samplers.py
• Please check the following directory for supported extensions.
  • py/defs/ext/

Asynchronous sending to Eagle (experimental)

This extension can send items to Eagle in a background thread instead of synchronously. This may reduce the time spent waiting for Eagle during image generation, at the cost of using an in-memory queue.

How to enable

Set the EAGLE_ASYNC_SEND environment variable to 1 before starting ComfyUI.

Examples:

• Linux / macOS (bash, zsh, etc.):

  export EAGLE_ASYNC_SEND=1
  python main.py
  

• Windows (Command Prompt):

  set EAGLE_ASYNC_SEND=1
  python main.py
  

• Windows (PowerShell):

  $env:EAGLE_ASYNC_SEND = "1"
  python .\main.py
  

To disable asynchronous sending, do not set EAGLE_ASYNC_SEND, or unset it / close the terminal and restart ComfyUI.

Behavior and limitations

• When EAGLE_ASYNC_SEND=1 is set, sending to Eagle is performed by a separate thread.
• Generated images can finish processing in ComfyUI while sending to Eagle continues in the background.
• The send queue is in-memory only. If Python/ComfyUI is forcibly terminated (e.g. crash, kill -9, power loss), items that are still queued may be lost and never reach Eagle.
• This feature is experimental and its behavior or availability may change in future versions.
• If you suspect a problem with sending to Eagle, try running again without EAGLE_ASYNC_SEND to fall back to synchronous sending.

Change History

• 2025/12/17 1.2.0 If the environment variable(EAGLE_ASYNC_SEND) is set, send to Eagle in a separate thread (experimental)
• 2025/11/09 1.1.9 Fixed a bug where prompts were not saved in ComfyUI v0.3.68
• 2025/11/09 1.1.8 Fixed a bug where prompts might not save. Fully supports workflows using loops.
• 2025/11/06 1.1.7 Bug Fix for Node Analysis in ComfyUI-Easy-Use
• 2025/11/05 1.1.6 Added IMAGE to output. Outputs the IMAGE received as input without modification.
• 2025/11/03 1.1.5 Fixed a bug that could cause the error "AttributeError: 'NoneType' object has no attribute 'caches'".
• 2025/11/02 1.1.4 Fixed a bug that caused errors in certain workflows
• 2025/11/01 1.1.3 Added the ability to set different values for each image in a batch by passing a list to each input in filename_prefix/custom_tag_pattern/eagle_folder/memo/extra_metadata.
• 2025/10/29 1.1.2 Fixed to correctly save the value for each loop iteration to metadata when called multiple times within a loop
• 2025/10/29 1.1.1 Add the size of the filepath (list) to the output
• 2025/10/29 1.1.0 Added support for supplying per-image positive/negative prompts via list inputs
• 2025/10/28 1.0.6 Changed to no preview for improved performance
• 2025/10/27 1.0.5 Supports Eagle API Token (specified via environment variable)
• 2025/10/26 1.0.4 Fixed cases where errors occur related to DynamicPrompt
• 2025/08/22 1.0.3 Fixed an issue where using Qwen-Image caused an error
• 2025/03/23 1.0.2 Fixed an error when using embeddings
• 2025/02/07 1.0.1 Support for comfyui-prompt-control
• 2025/01/04 1.0.0 First release

Acknowledgements

• The implementation for metadata acquisition is based on ComfyUI-SaveImageWithMetaData
• The implementation for sending to Eagle is based on D2 Send Eagle
• The tag customization specification is based on the WebUI extension Eagle-pnginfo
• Thanks to nkchocoai, da2el, and bbc_mc for creating and publishing such wonderful programs