ComfyUI Image Checker
Inspect ComfyUI image exports for embedded node graphs, execution JSON, prompts, LoRA configurations, and generation seeds in browser RAM.
ComfyUI Workflow Data:
ComfyUI saves your full execution graph and UI canvas directly inside PNG text chunks. Anyone with the file can drag it into ComfyUI to inspect your exact pipeline, models, and prompts. Inspect or sanitize files before sharing.
Drop your image here to check AI metadata
Drag & drop a JPG, PNG, or WebP file (or up to 30 images for batch checking), or click to browse.
How to use the ComfyUI Image Checker
Checks ComfyUI PNGs for the embedded workflow JSON that lets anyone reload your whole node graph.
- Step 1: Drop a JPG, PNG or WebP file, or up to 30 images at once for a batch check. The scan starts straight away and runs in your browser.
- Step 2: Read the Detailed Signal Report. Each signal is marked Detected, Not detected, Not assessed or Unsupported.
- Step 3: Open the workflow schema to see whether the full node graph and prompt are embedded.
- Step 4: Optional: click “Clean Supported Metadata”. The cleaned copy is checked again before you use “Download Clean Image”.
What you'll see
- Whether workflow and prompt JSON are embedded, with model, seed and prompt values when present.
- This page uses the same analysis as the AI Image Checker. What changes here is the explanation for this platform or generator, not the scan itself.
Understanding ComfyUI Embedded Metadata
Unlike traditional image generators that record only a flat text prompt, ComfyUI serializes its entire modular node architecture into the output file. While this makes sharing and reproducing open-source workflows seamless, it also creates privacy and intellectual property concerns:
Serialized API prompt containing sampler seeds, CFG guidance, step counts, and conditioning text.
Complete canvas topology, node coordinates, custom group labels, and wire links between nodes.
Local folder paths to checkpoint safetensors, custom LoRA weights, and personal directory structures.
Related Tools & Guides
Frequently Asked Questions
What exact metadata does ComfyUI embed in exported PNG images?
ComfyUI embeds two distinct JSON structures inside standard PNG text chunks: the 'prompt' chunk and the 'workflow' chunk. The 'prompt' chunk contains the raw API execution graph sent to the generation queue (including node classes like KSampler, CLIPTextEncode, and VAEDecode). The 'workflow' chunk contains the complete front-end visual canvas state, including node placement coordinates, pin-to-pin wiring connections, custom node titles, and bypass flags.
Can ComfyUI metadata leak private local file paths or usernames?
Yes. Because ComfyUI records model loader configurations, the serialized node graph frequently includes full filesystem paths to local checkpoint models, LoRAs, VAEs, or output directories (e.g., 'C:\Users\username\ComfyUI\models\checkpoints\...'). Inspecting and stripping this metadata prevents accidental disclosure of internal directory structures and personal user accounts.
Can someone recreate my generation pipeline simply by downloading my image?
Yes. ComfyUI natively supports drag-and-drop workflow reconstruction. When an uncleaned ComfyUI image is dropped onto the canvas, the application automatically deserializes the 'workflow' chunk and restores every node, slider setting, seed, and prompt connection exactly as you configured it.
Does converting ComfyUI PNGs to WebP or JPEG automatically remove the workflow?
Standard web conversions often drop ancillary PNG chunks, but re-encoding introduces compression artifacts and quality loss. Our lossless cleaner strips the ComfyUI 'prompt' and 'workflow' chunks directly at the container level, preserving the original pixel data, color profiles, and visual fidelity with zero re-compression.
How does our browser-side checker inspect ComfyUI files safely?
Our inspection engine reads the PNG chunk headers directly inside browser RAM without sending bytes over the network or executing untrusted workflow scripts. It parses the JSON headers in memory to display the embedded prompt and model details, and allows one-click lossless stripping if you choose to sanitize the file.
When to use it
- You want to share a ComfyUI image without giving away your workflow.
- You want to check a batch of outputs before uploading them.
When another tool fits better
- You want to remove the workflow:AI Metadata Remover
- You want to read the prompt and settings:AI Metadata Viewer
What the ComfyUI Image Checker cannot do
- It reads metadata only. It does not analyse pixels, so it cannot tell whether an image was made by AI, and pixel watermarks such as SynthID are outside its reach.
- “Not detected” means none of the supported fields were found. Platforms and screenshots often strip metadata, so a clean result is not proof of anything.
Which tool do I need?
Every tool runs in your browser on the same engine. Pick the one that matches what you want to do.
| I want to… | Use |
|---|---|
| Remove AI prompts, EXIF, GPS and Content Credentials in one go | AI Metadata Remover |
| Remove camera details (make, model, dates) | EXIF Remover |
| Check for and remove location data | GPS Location Remover |
| Remove only C2PA Content Credentials | C2PA Remover |
| Clean author and software details from a PDF | PDF Metadata Remover |
| See everything a file contains, change nothing | Image Metadata Viewer |
| Read the prompt, seed and model of an AI image | AI Metadata Viewer |
| Check many images for AI metadata at once | AI Image Checker |
| Audit every AI field in technical detail | AI Metadata Checker |
| Check who signed an image’s Content Credentials | C2PA Checker |
| See what a photo reveals about you | Image Privacy Checker |
| Change metadata instead of deleting it | Metadata Editor |