PdfCraft is an open-source PDF reader and editor, written from scratch in pure Rust, that runs as a native app on macOS, Windows, Linux and FreeBSD, and also on the web. It is for people who work with documents they would rather not upload: contracts, scanned paperwork, signed forms, anything where "it syncs to our cloud" is the wrong answer. The project is new, created at the end of September 2026, and has already gathered around 12,100 stars and 4,600 forks.
What it does
Its authors call it "the PDF workbench", a clean-room reimplementation of Adobe Acrobat. In practice that means the ordinary paperwork jobs:
- read and page through documents, with a tools panel on one side and threaded comments on the other (the README screenshot shows a 13-page specimen file open with 20 comment threads)
- organize pages, including reordering a contract by dragging pages around like cards on a table
- combine and split files
- secure documents
- OCR for scanned pages, with a caveat covered below
Two things it deliberately skips: an account and telemetry. Nothing in the material describes a sign-in or a server component, and the project's claim is that nothing leaves your machine.
How it works
The whole stack is Rust and the project's own code: its own PDF parsing and rendering rather than a wrapper around an existing engine. Clean room, as the authors use the term, means written without reusing Acrobat's internals. It ships in three pieces, all open source: the engine, a command-line tool and the app, so document operations are scriptable from a terminal and not only clickable.
Editing is non-destructive. After you save, the original bytes stay untouched, and undo reaches back through the edit history rather than one step.
The headline reliability figure, 983 files with 0 crashes, is the project's own test suite and the authors' own claim. No independent crash or compatibility measurement appears in the material, so read it as "their harness is green", not as a verified compatibility score.
Getting started
The code is at github.com/storytold/pdfcraft, and the project keeps a page at getartcraft.com/apps/pdfcraft as one of several ArtCraft apps, with a community Discord at discord.gg/artcraft. Platform support claimed by the README: macOS, Windows, Linux, FreeBSD and the web.
Cost: the app is free and open source (GitHub reports Apache-2.0, while the README badges say MIT or Apache-2.0). No API key, paid cloud service or subscription appears anywhere in the material, OCR included: the OCR model is something you download separately and run on your own hardware. The material does not name the model, its size or where to fetch it.
What it needs from your machine is not stated: the project lists no hardware requirements, so there is no published minimum memory, GPU or chip. The saved README is cut before any install section, so there is no exact command to quote here. Check the repository itself for prebuilt binaries or the build steps, which for a Rust project of this kind means a clone plus the standard Rust toolchain.
When to use it / when not
The main catch is OCR: searchable text from scans works only once you have fetched a model yourself, so a fresh install will not read a scanned page out of the box. The second is age. This repository is about two weeks old with commits still landing daily, which is early for software whose job is to not corrupt a legal document.
Use it instead of Adobe Acrobat when the files are sensitive, you want the page tools running locally, and you would rather not sign in to or pay for a desktop suite. Use it instead of a browser PDF tab when you need to reorder, merge or annotate rather than just look.
Don't use it if you depend on Acrobat features the material never claims, such as forms workflows, certified digital signatures or prepress checks, or if you cannot risk subtle breakage from a brand-new PDF implementation on your most important files.
Alternatives
The reference point is Acrobat, which PdfCraft names directly. For viewing alone, Preview on macOS and the viewers built into Chrome and Firefox already cover most needs and cost nothing. Against the web-based "merge and split your PDF here" sites, the difference is where the file goes: those upload it, PdfCraft does not. The material makes no comparison with other open-source PDF editors or libraries, so this entry cannot rank it against them.
Anyone who handles PDFs regularly and dislikes the cloud default should watch this one closely. It is native on five targets, free, and split into an engine, a CLI and an app, which is the right shape for both scripting and daily use. Treat the 983-file clean run as the authors' own evidence, keep backups of anything you cannot re-create, download the OCR model before you need it, and give the project a few releases to settle before it takes over contract work.