User guide
From prompt list to organized outputs.
Higgsfield Automation turns prompts into a queue and runs them through the Higgsfield page open in your browser. You choose the model and generation settings on higgsfield.ai; the extension handles the repeated queue work.
Choose a workflow
Runs one or more text prompts through the Higgsfield video workspace.
Uses uploaded start frames, with optional end frames, for video prompts.
Runs text prompts through the Higgsfield image workspace and applies the selected image batch size.
Pairs uploaded reference images with prompts in the order shown in the image rack.
Add prompts
Type prompts
Enter one prompt per blank-line-separated block. A prompt may span several lines; a completely blank line starts the next prompt.
Import prompts
Use the import controls above the prompt field:
- TXT files are split into prompt blocks.
- CSV, XLS, and XLSX files open a preview where you can choose the sheet and prompt column.
- The importer suggests the column containing the most usable text.
- Choose Replace to start a new list or Append to add to the prompts already present.
Configure the batch
Outputs per prompt
| Workflow | Behavior |
|---|---|
| Image workflows | The extension adjusts Higgsfield's image Batch Size control and submits each prompt once. |
| Video workflows | The extension submits the video prompt the selected number of times. |
Concurrency and pacing
- Concurrent prompts controls how many generations may be resolving at the same time.
- Delay range adds randomized pacing between submissions.
- Wait limits set ceilings for detecting a started generation and waiting for it to finish. A generation continues immediately when it completes; these values are maximums.
Retries
If a generation does not start or cannot be resolved, the extension retries according to your settings. The queue and report distinguish submission retries from result or download retries.
Use reference images
Pair images manually
Upload images, drag them into the intended order, and review the prompt-to-image preview before starting. Batch validation warns when a workflow is missing required images.
Match character images automatically
In Image to Image mode, name uploaded files after the characters they represent. When enabled, a prompt that mentions the filename's character name receives that image as a reference.
Chain from the previous output
Start a prompt with >> to use the previous output as the next prompt's reference:
- For video, the previous result can provide the next start frame.
- For image workflows, the previous image becomes the next reference.
Downloads and reports
- Auto-download saves results after they finish.
- Folder name places files in a subfolder of your browser's Downloads directory.
- Filename mode can use prompt-based names or retain the original source name, depending on the selected option.
- CSV report records prompt status, retry stages, output details, filenames, and completion information.
Stop and resume
Cancel stops new work and preserves a compatible resume point. Resume is offered only while the prompt, image, mode, and relevant settings signature still matches the interrupted batch. Editing that batch starts a fresh run.
Troubleshooting
- Keep the intended Higgsfield workspace open and signed in.
- If the extension cannot reach the tab, refresh the Higgsfield page and try again.
- If selector configuration cannot load, make sure network filters allow
higgsfield-automation.pages.dev. The selector endpoint uses an extension request header, so a normal browser tab may not show the JSON directly. - Review the action log for the failed prompt and retry stage before restarting a batch.