Every tool on ToolGarden starts from one premise: if the browser can do the work itself, the file does not go to a server. JSON parsing, image codecs, PDF page reassembly, audio transcoding, OCR and background-removal models all run in-page now, via Canvas, Web Workers, WebAssembly and local models.

The practical consequence is that you can drop an API response containing tokens, an unsigned contract, a scanned ID or an unpublished manuscript straight into a tool, because none of it leaves this device.

Choosing between these tools

Find it by data type: The top navigation groups tools by JSON, image, audio, PDF, text, file merging, encoding, QR codes and subtitles. When you know what kind of file you are holding, the category page is the fastest way in.
Find it by action: Within a category, tools are named after the action; convert, compress, merge, split, diff, validate, extract. Search matches action words directly, so typing "merge" lists the merge tool for every format.
Not sure which one applies: Every category page carries a "choosing between these tools" section and a format comparison table, and every tool page has steps, a worked input/output example and its limits. Reading the category page first saves redoing the job after picking the wrong format.
Curious how it works: The blog covers in-browser media processing, model inference and format specifications, and tool pages link down to the relevant deep-dive guide.

Formats and capabilities compared

Local processing differs from upload-based tools in more than privacy; the capability limits differ too, which is what tells you when to reach for desktop software instead.

Dimension	Processed in the browser	Uploaded to a server
Where data goes	The file stays on your device; only code and models come down the wire	The file must be uploaded, after which retention and deletion are the operator's policy
File size	Bounded by browser memory; very large files can fail	Bounded by server limits, and usually handles more
Speed	Depends on your device; transcoding is fast, model inference is slow	Depends on server capacity, and is usually more predictable
Works offline	Most tools work offline once their assets are cached	Requires a connection

Limits that apply across this category

Every tool is free and needs no account. The site does count page visits to learn which tools get used; that is a separate data flow from what you put into a tool, and the privacy policy spells out the distinction.

The trade-off for local processing is that the compute and the memory are yours. Files up to tens of megabytes are generally fine, but very long media, PDFs with a great many pages, and AI tools that download tens of megabytes of model can be slow or fail outright on modest hardware or a weak connection. Desktop or command-line software is the better choice there.

Most conversions are lossy or not fully reversible. Always keep the original file, and check the output before a converted result goes into a real process.

Topic guides

Start with a complete guide, then use the focused articles for a specific workflow or question.

Complete guide

ToolGarden

Popular tools

JSON tools(22)

Image tools(22)

PDF tools(9)

Audio tools(15)

Text tools(5)

File merge tools(9)

Info codec(1)

QR tools(2)

Subtitle tools(1)

Other tools(4)

What this category is for

Choosing between these tools

Formats and capabilities compared

Limits that apply across this category

Topic guides

The Complete Guide to Browser-Local Data Processing

JSON Tools Master Guide

Image Processing Without Upload

PDF Tools Guide

QR Code & Subtitle Tools Guide