ComfyUI Auto Tagger for Eagle

ComfyUI Auto Tagger is a plugin for Eagle that automatically extracts metadata from AI-generated images. It provides two main features:

1. Generation Info Inspector — Instantly view prompts, seeds, and generation parameters when you select an image (with copy buttons for quick reuse)
2. Auto-Tagging — Save extracted metadata as Tags and Notes for organization and filtering

It supports ComfyUI, Stable Diffusion WebUI (including Automatic1111, Forge, and other variants; referred to as "A1111" below), and Civitai generated images in PNG, WebP, and JPEG formats, and allows you to filter which information to import.

[Features]

Generation Info Inspector: When you select an image in Eagle, a panel automatically appears showing the generation metadata — prompts, seed, sampler, model, and LoRA — with a copy button for each field. Multi-sampler workflows display as tabs (Base / Step 2, etc.).

• Multi-Format Support: Supports ComfyUI workflows (including complex multi-sampler), A1111 (Stable Diffusion WebUI) parameters, and Civitai generation metadata.
• Metadata Extraction: Automatically extracts the models used (Checkpoints, LoRA), prompts, and generation settings (Seed, Sampler, Steps, CFG).
• Flexible Output:
  • Tags: Adds extracted info to Eagle tags (e.g., #checkpoint_name, #lora_name, seed:12345).
  • Notes (Annotation): Saves full prompts and parameters in the Note section for easy reference.
• Selective Import: Allows toggling specific items (e.g., "Import Checkpoint but ignore Seed") via checkboxes.
• Batch Processing: Efficiently processes multiple images with a progress bar.
• Advanced Workflow Analysis: Dynamically analyzes ComfyUI workflows to trace the actual execution path. Accurately extracts parameters even from complex workflows with multiple generation stages (HiresFix, FaceDetailer, etc.).
• Suspicious Node Detection: Detects nodes with missing required inputs in ComfyUI workflows and allows users to decide whether to include or exclude them from metadata extraction.
• Force Delete Mode: Removes all tags and notes from selected items without analysis (Shift + Click on "Delete Info").
• Debug Mode: Detailed logs for troubleshooting (toggle via checkbox).
• No External Dependencies: No additional libraries required — works out of the box after installation.
• Utility: Provides a dedicated button to safely remove only the tags/notes added by this plugin.

[Ultimate Automation: Eagle Metadata Bridge]

Eagle Metadata Bridge is a ComfyUI custom node that replaces the standard Save Image node. Every time you generate an image, it automatically:

• Sends the image to Eagle
• Attaches tags (checkpoint, LoRA, prompt tokens, seed, sampler…) and a structured annotation

When used alongside this plugin, tag management becomes even more powerful — images arrive in Eagle already tagged, and you can re-run this plugin at any time to update, filter, or batch-process them further. The node embeds an identifier into each saved image, allowing this plugin to reliably retrieve the prompt and generation parameters by tracing the exact workflow graph instead of relying on heuristics or inference.

[Advanced Workflow Trace Logic]

The plugin traces complex ComfyUI workflows to identify the actual generation pipeline, supporting multi-stage setups with multiple samplers and refinement stages (HiresFix, FaceDetailer, etc.).

• Path-Based Extraction: It traces the execution path from the output nodes back to the source. This ensures that in "All-in-One" workflows, only the parameters that actually contributed to the image are recorded.

• Multi-Stage Support: Captures data from every generation step, including KSamplers, FaceDetailers, and HiresFix, providing a complete "recipe" for reproduction.

• Accurate Parameter Retrieval: Even with complex custom nodes like "Lora Loader Stack" or Reroute/Primitive nodes, the logic recursively resolves values to ensure accurate retrieval.
  

• Full Audit in Notes: Every detected generation step is recorded in detail within the Eagle Notes section, providing a precise "recipe" for reproduction.

[Heuristic Node Detection]

ComfyUI metadata doesn't record the actual execution path. The plugin traces back from the final image save node to extract prompts and parameters. In this process, nodes that weren't actually used may be included.

For example, a workflow designed for txt2img (text-to-image) might contain img2img (image-to-image) nodes. If you used only txt2img, those img2img nodes were never executed — but the plugin cannot detect this automatically. When suspicious nodes are detected, the plugin displays a dialog, allowing you to decide whether to include or exclude them from metadata extraction.

You can configure how to handle suspicious nodes in the settings (see Configuration section below).

[Settings]

You have full control over how your library is organized:

• Extraction Targets: Selectively import Checkpoints, LoRAs, Prompts, Seeds, Steps, CFG, etc.

• Output Destination: Toggle Tags and Notes independently to match your workflow.

The plugin provides several configuration options accessible via the settings dialog (gear icon):

• Processing Settings

  • Process Items per Batch: Set the number of images to process at once (default: 5). Processing too many images simultaneously may consume excessive memory.

  Suspicious Node Handling

  • Suspicious Node Handling: Configure how to handle suspicious nodes (nodes with missing required inputs):
    • Exclude: Automatically exclude suspicious nodes from metadata extraction
    • Ask: Show a dialog for each suspicious node (recommended)
    • Include: Include all nodes regardless of missing inputs

  Dictionary Settings

  • Fetch dictionary from online (Recommended): Fetches the latest custom node definitions from GitHub. If disabled, the bundled dictionary will be used. Changes take effect on next startup.

  Tag Generation Settings

  • Include parameters from all samplers in tags: When enabled, parameters from all samplers in the workflow are added to tags. When disabled (default), only parameters from the first executed sampler are added. Enable this if you want to record all parameters from complex workflows with multiple refinement stages (HiresFix, upscale, etc.).

  Debug Settings

  • Enable debug mode: Outputs detailed debug logs to console. Useful for diagnosing issues or reporting bugs.

  Cache Settings

  • Do not use cache: When enabled, the plugin will not cache metadata for images. This is useful for testing or when you want to ensure fresh metadata is always extracted. Changes take effect immediately.
  • Clear All Cache: Permanently deletes all cached metadata files. This frees up disk space and allows you to start fresh.
  • Delete Cache (Inspector): When viewing an image in the Inspector panel, a delete button (🗑️) appears at the bottom. Click it to delete the cached metadata for that specific image.

  Cache files are stored in {Library Path}/.eagle/plugins/comfyui-auto-tagger/metadata-cache/ and are automatically used to speed up repeated metadata viewing.

[How to Use]

1. Select one or more AI images in Eagle.

2. Right-click and choose Plugins > ComfyUI Auto Tagger.

3. Configure your settings (First-time setup recommended).

4. Click Start Tagging.