The payload boundary is the browser.
jsontools.me is designed so the service can deliver tools without receiving the data used in them.
Document processing
Input bytes are transferred to a dedicated Web Worker and parsed by the json-buf WebAssembly module. Tool operations use an in-memory document handle. There is no application endpoint for uploading JSON payloads.
Opening a focused tool page, installing the PWA, or using the service offline does not change that boundary.
Storage is off by default
The current document is held in memory for the active session. “Remember this workspace on this device” is an explicit option. When enabled, the document is written only to browser storage on that device. Turning the option off removes the stored workspace.
Preferences such as theme and parser mode may be stored locally without storing the document itself.
Measurement without payloads
The analytics event schema allows coarse metadata such as tool name, input-size bucket, duration bucket, runtime, transition, copy/export action, and PWA lifecycle. It rejects payload, input, output, value, path, and query fields.
No analytics collector is required for local serving. A deployment can connect the typed metadata stream to a privacy-conscious collector without changing engine behavior.
Threat-model limits
A client-side architecture reduces the service’s access to payloads; it does not make an already compromised browser, extension, operating system, or device safe. Sensitive-data detection is a review aid, not a compliance guarantee.
Before using production secrets, inspect the runtime diagnostics, browser extensions, and the application source you are running.
Network requests you should expect
On first use, the browser requests HTML, styles, JavaScript, the worker, the WebAssembly binary, icons, and fonts if packaged. The service worker caches application assets for offline use. Tool operations do not create a document request.