Run several PDF tools in a row

Most real work with documents is more than one operation. Rotate the scans, then compress them. Delete the cover page, then add page numbers, then watermark the result. Number a production set and lay it out four to a sheet. Doing that one tool at a time means downloading a file, finding it, uploading it to the next tool, and repeating that for every document you have.

This does the whole sequence in one pass, over as many files as you like, without any of them leaving your machine.

How it works

Choose your files, then pick up to four operations in the order you want them. Each step gets its own controls, and each one runs on the result of the step before it — so step two sees what step one produced, not the original.

Files are processed one at a time rather than all at once. Ten twenty-megabyte documents decoded simultaneously will end a browser tab on a phone, and the whole point of handing over a stack is that you can stop watching it.

When something goes wrong

A file that fails at step three keeps the work of steps one and two, and the result says which step it stopped at and why. “Compression failed at step 2 of 3” is something you can act on; “this file failed” is not.

One file failing does not stop the others. If nine documents go through and one is corrupt, you get nine finished files and one row explaining the tenth, rather than an error page and nothing to show for it.

Why only some tools appear

A step has to take one file, produce one file, and know what to do without being shown where. That rules out three kinds of tool.

Anything that needs a visual placement — signing, redaction, freehand editing — needs you to point at a spot on a specific page, which cannot be answered in advance for a stack of documents you have not looked at.

Anything that produces a report rather than a document, such as inspection or metadata reading, has nothing for the next step to work on.

And anything that takes several files and makes one, such as merging, is not a link in a chain over each file.

Everything else is offered, and the list comes from the same description of each tool that the tool’s own page uses. A step in a chain and a visit to the tool page send exactly the same request to exactly the same code, so a tool cannot behave one way here and another way there.

Nothing leaves the device

Every step runs in a worker inside your browser. The file passes from one step to the next in memory, never touching a server, never being written anywhere you did not ask for. That is the part a chain of separate online tools cannot offer: even if you trust each one, the document has been uploaded three times by the time it reaches the end.