Free · open source · works offline
Seven office apps. One download. No account, ever.
Mail, documents, spreadsheets, slides, pictures, image editing and video — on an engine we wrote ourselves, reading the Microsoft formats and the mail archives you already have. Unplug the machine and all of it still works.
Windows today; macOS and Linux build from source. Open source under the AGPL, with a commercial licence for anyone who needs one.
7
apps in one installer
0
accounts, sign-ins or telemetry
27
file formats read or written
577
engine tests, all passing
What you get
The whole day’s work, in one window
Each app opens in the same shell, with the same design and the same shortcuts. Nothing here is a web page in a frame.
Every account, every archive, one inbox
IMAP and SMTP for the accounts you use now, and — the part no other free client does properly — the archives other clients leave behind. Outlook .pst and .ost are read in full: folders, messages, recipients and attachments, out of the node and block B-trees rather than through Outlook.
IMAP · SMTP · .pst · .ost · .olm · mbox · .eml · .emlx · .msg
Word
Documents that open the same everywhere
A word processor on a .docx engine we wrote, not a converter wrapped around someone else’s. Tracked changes, comments, styles, sections and pagination — and a PDF writer with no dependencies underneath the export.
Reads and writes .docx · reads .odt, .rtf, .doc, Markdown, text · exports PDF
Worksheets
Real formulas, real recalculation
A spreadsheet with an actual calculation engine — parser, evaluator and dependency ordering — so a workbook recalculates in the right order rather than showing you the last values it was saved with. Number formats, merged cells, charts and conditional appearance all render.
Reads and writes .xlsx and .csv · reads .ods, .xls, .tsv
Presentation
Slides that survive the round trip
A .pptx engine that reads the package, resolves layout and master inheritance, and renders the scene — so a deck that came from PowerPoint still looks like the deck that came from PowerPoint after you have edited a title on it.
Reads and writes .pptx · reads .odp, .ppt
Pictures
A viewer that opens before you blink
The double-click app: every common image format, EXIF read and orientation honoured, folder browsing, and PDFs in the same window. Nothing to wait for and nothing to configure.
Every common image format · EXIF · PDF
Image
Crop, correct, annotate, export
A non-destructive editor: the edits are a pipeline over your file, and your original is never written to. Export when you are finished, to the format and size you actually need.
Exports .png, .jpg, .webp
Video
Trim and export without a render farm
Container probing and an edit timeline for the everyday job — trim the ends off, cut the middle out, export. No ffmpeg to install, no native binaries, nothing that needs a toolchain.
Reads what the browser engine decodes · exports .webm
Why this exists
Most free office software asks you to give something up
Your file format, your privacy, your old mail, or your patience. This asks for none of those.
It opens what you already have
Microsoft formats are the point, not an afterthought. So are the mail archives you have been carrying between computers for fifteen years — .pst, .ost, .olm, mbox — which most clients simply cannot read.
It works offline
No account, no sign-in, no telemetry. There is exactly one network call in the whole suite and it is the update check, which is a switch you can turn off. Everything else runs on your machine.
It does not damage your files
The engine rewrites only the parts of a document it deliberately edited. Charts, pivot caches, macros, signatures and embedded media are returned byte-for-byte as they arrived.
The source is published
Under the AGPL, which means you can read it, change it and share it — and check every claim on this page against the code rather than taking our word for it.
The engine
Two decisions shape everything else
They are worth stating out loud, because they are the reason fidelity is high on day one instead of after five years of bug reports.
Documents live in the backend
The OOXML engine inflates with zlib and works in buffers — it is a Node engine, not a browser one. A document opens, edits and saves in the main process, and the window draws the view model it is sent. A spreadsheet asks for the viewport it is about to paint and nothing else, which is why the size of a workbook has no bearing on the size of a window.
We only rewrite what we edit
An .xlsx is a zip of XML parts, and almost everything we will not implement for years lives in a part that can be carried through untouched. So fidelity degrades only in the specific places we choose to touch — which are also the places we test.
The part nobody else does
Fifteen years of Outlook mail, opened without Outlook
.pst and .ost are read in full — the node and block B-trees, the heap, property and table contexts, folders, messages, recipients and attachments. Verified against a real 906 MB .ost: 467 folders and 12,172 messages indexed in under a second.
That work found three things no amount of reading the specification would have given — including that Outlook 2013 and later compress the larger blocks of an .ost with zlib, which is not in the published structure and which a reader that does not notice sees as an archive full of empty messages.
Questions
Frequently asked
Is it really free?
Yes — free to download, free to use, for anything, including at work. It is published under the GNU AGPL v3.0. A separate commercial licence exists for the specific case of embedding it in a closed-source product or running a modified version as a hosted service, which is a conversation rather than a price list.
What is the catch?
There is no upsell inside the application, no account, no cloud tier and no file limit. Rutba sells business software at rutba.io; this suite exists because the engine underneath it was built for that platform anyway, and a desktop suite is a better use of it than a drawer.
Will it break my .docx or .xlsx?
It is designed not to. The engine rewrites only the parts it edited and returns everything else byte-for-byte — charts, pivot caches, macros, signatures, embedded media. That said: it is version 1.0. Keep a copy of anything irreplaceable, as you would with any new tool.
Does it send anything anywhere?
One HTTPS request to GitHub for the release feed, so the app can tell you a new version exists. It carries nothing about you, your documents or your machine, and it can be switched off, after which no request is made at all. Nothing else in the suite contacts anything.
How do I report a bug or ask for a feature?
On GitHub. The issue tracker is the support desk for this product and the place the roadmap is argued out in public — see the community page.
Download it and see
Windows, macOS and Linux. No sign-up, no email address, no trial period.