What works today
PdfCraft's README describes it as a PDF workbench, and its list of what is good today covers most everyday PDF work: viewing and search; organizing, combining and splitting; most kinds of comment; filling and authoring forms; passwords, redaction and sanitizing; basic digital signatures; printing; an Accessibility Checker; and control by scripts and AI agents.
Pages render with kerning, ligatures, right-to-left and complex scripts, vertical CJK and colour emoji, and large pages render in tiles so text stays sharp. Organize pages shows every page at once for rotating, deleting, inserting and moving, and Combine, Extract and Split keep links, form fields, layers and attachments intact. Encrypted files open with their passwords, the author's restrictions are honoured, and edits are saved encrypted under the same keys.
How it compares with Acrobat
The project measures itself against Adobe Acrobat Pro. In October 2026 the README put PdfCraft at alpha, with 57% of tracked features and 94% of the must-haves in some form, and about 55% of the way to replacing Acrobat for everyday work. Those are the project's own estimates, and it says quality has not yet been compared with Acrobat side by side.
The README is direct about the gaps. Editing existing text is not yet reliable, especially in CJK and right-to-left scripts. There is no Office-to-PDF, and Word export loses layout. OCR reads only the Latin alphabet without accents, so Chinese scans can't be recognized. Pages are still drawn by the hayro crate while PdfCraft's own renderer is built, and fuzzing still finds crashes on hostile files. For Chinese readers there is one more: release builds can show Chinese interface labels as empty boxes.
Privacy and agents
The README says documents never leave your machine, with no account, no telemetry and no cloud processing. The pdfcraft-cli tool and the MCP server for AI agents are opt-in: the server starts only when an agent launches it, opens no network port, and can be confined to one folder with --root. A separate control channel for driving the app is off by default and listens only on loopback with a token.
PdfCraft on GitHub, and two projects with similar names
Three open-source projects on GitHub have almost the same name, and none of their READMEs mentions the others. When we checked them on 17 October 2026:
- PdfCraft, the app on this page, at storytold/pdfcraft: the ArtCraft team's PDF workbench, a native app written in Rust for Windows, macOS, Linux and FreeBSD, with a WebAssembly build that runs in the browser. MIT or Apache-2.0; on GitHub since 30 September 2026.
- PDFCraft at PDFCraftTool/pdfcraft: a PDF toolkit that runs in the browser, built with Next.js and WebAssembly; its repository description mentions 90+ tools. It has its own website, pdfcraft.devtoolcafe.com, and its releases include desktop builds for Windows, macOS (Apple silicon) and Linux. AGPL-3.0; on GitHub since January 2026.
- PDF Craft at oomol-lab/pdf-craft, from OOMOL Lab: a Python library that uses OCR models to turn scanned PDFs, such as books and papers, into Markdown or EPUB, and can translate them. Its online service is at pdfcraft.ai. MIT; on GitHub since February 2025.
The download box on this page links only to files published in storytold/pdfcraft.
About the Craft Apps
PdfCraft is one of the Crafting Apps from the ArtCraft team: free, open-source tools written from scratch in Rust, each able to stand on its own, alongside DeckCraft for presentations, PhotoCraft for images and DesignCraft for page layout. Its development notes mention migrating settings from PrintCraft, an earlier name. The code is dual-licensed under MIT or Apache-2.0; the ArtCraft name and logos are trademarks outside that license. PdfCraft is not affiliated with Adobe.