Published by PitchChanger · Updated · How these guides are made
How to use this workflow
Capture a clean baseline
Open developer tools before loading the tool. Preserve the Network log if navigation is involved. Load the page without choosing a file and record the origins already contacted.
Add one action per checkpoint
Select a harmless file, preview, process and save in separate steps. Inspect new requests after each action: method, destination, resource type, payload and initiator. Clear filters before concluding there were no requests.
A request classification worksheet
This original proposed log avoids classifying a request by size alone. A large GET can be a model download; a small POST can contain identifying metadata. A blob: URL normally identifies a browser-managed object and is not itself evidence of a remote upload.
Swipe or scroll sideways to read all columns.
| Request clue | Inspect next | Conclusion limit |
|---|---|---|
| GET model asset | Response and initiating script | Incoming asset, not source upload by itself |
| POST after file selection | Payload and destination | Could carry media or metadata; inspect |
| No visible new request | Filters, frames and observation window | No request observed in this capture |
Verify the result and record its limits
Repeat with a different harmless filename and compare payloads if you are checking filename transmission. Do not publish a raw archive containing private headers or tokens. A single capture describes the browser, feature and observation window tested; it does not certify all future versions or extension behavior. Keep privacy claims proportional to that scope.
Common questions
Was this example measured on my device?
No. The worksheet and any hypothetical numbers are explicitly labeled. Use a harmless recording to collect your own observations and record the browser, date and exact operation.