SoloPDF
v0.8.0 · MIT · 5 platforms · notarized · zero telemetry

A PDF reader that behaves. Free.

Free · ad-free · local-first PDF reader — highlights become notes

Download for macOS (Apple Silicon / Intel) All platforms
No ads, no pop-upsNo accountNothing uploadedMIT open sourceUpdate check off by default
★ – stars · ⬇ – downloads

SoloPDF in 60 seconds

  • Highlights become notes. Highlight a passage and SoloPDF writes <document>.annotations.md next to the PDF — an ordinary Markdown file any editor can open and git can track.
  • Not just PDF. EPUB / MOBI / AZW3 reflow, CBZ / CBR comics, DjVu scans, FB2 e-books, multi-page TIFF, and a TXT book mode.
  • On-device OCR. Turn scans into searchable PDFs or Markdown, fully offline — system Vision on macOS/iOS, bundled PP-OCRv4 on Windows/Linux.
  • Built for papers: links work with Back/Forward, hover previews for Figure / Table / [12] references, split view, ink and undo, and selection translation on the system's on-device engine — no network.
  • Document tools. Merge, split, reorder, compress, encrypt/decrypt, sign — every operation writes a new file and leaves the original untouched.
  • Five platforms. macOS (universal, Intel + Apple Silicon), Windows x64 and ARM64, Linux x64 and ARM64, plus iOS and Android.
  • CLI and MCP included. Same rendering engine as the app, scriptable and drivable by AI agents out of the box.

Highlights become notes

Swipe a highlight in any PDF and SoloPDF writes document.annotations.md next to it — a plain Markdown file:

# Annotations of paper.pdf

## p.23 — Highlight
> the highlighted source text
your note — edit it in any editor
[Jump to source](solopdf://open?file=…&page=23&annot=a1b2c3)

Open it with SoloMD: editable, full-text searchable, git-versionable. "Jump to source" takes you back to the exact spot in the PDF. After the PDF changes, highlights relocate by text fingerprint — and even when they can’t, they only grey out. Your notes are never lost.

Everything you need, nothing you don’t

Reading, annotating, document surgery — everything the mainstream readers have, and none of what they add.

⚡ Huge files open instantly

A 157MB / 1,048-page scan cold-starts in ~1s. Streaming loads + virtual scrolling, constant memory.

🔄 Navigate and lay out

Links work, with Back and Forward; hover to preview a Figure, Table or [12] reference; split view on one document; full screen and presentation mode; paper colours and printed page numbers; rotation, facing pages, auto-crop, auto-scroll.

✍️ Marks and ink

Highlight, underline, strikethrough, squiggly, sticky notes and region snapshots — plus pen, text boxes, rectangles, ellipses and arrows (stylus pressure supported), all undoable. Export a copy with standard PDF annotations that Preview and Acrobat show.

📝 Fillable forms

Type into AcroForm fields, tick checkboxes, export a filled copy — the original stays untouched.

🛠 Document tools

Rotate, delete, extract and reorder pages; merge and split; PDF ↔ images; compress; handwritten signatures and date stamps; set or remove an open password. Every operation writes a new file — the source is read-only.

📚 Shelf · dictionary · translate · speech

Translate any selection — on macOS / iOS with the system's on-device engine, no network. A cover grid with reading progress and search across every document. CC-CEDICT ships built in (120k entries, offline). System speech reads sentence by sentence and turns pages for you. Reading stats never leave the device.

🔍 CJK search that works

NFKC normalization — even the Kangxi-radical codepoints in Chrome-exported PDFs are handled.

🌙 Smart dark mode

Text pages invert for your eyes; pages with images stay as-is. No photo negatives.

🔒 Encrypted PDFs

Passwords remembered per-session only; optional privacy mode keeps excerpts out of your notes.

📖 Not just PDF

EPUB / MOBI / AZW3 reflow like an e-book — four backgrounds, three font stacks. CBZ / CBR comics get a dedicated reader with right-to-left paging. DjVu scans, FB2 e-books and multi-page TIFF open too. Highlights still become Markdown notes.

🔠 On-device OCR

Turn scans into searchable PDFs or Markdown. Photo capture gets perspective correction and deskew; two-column layouts export in reading order. Everything runs locally.

🚫 Never included

Ads, popups, subscription nags, watermarks, telemetry. Update check is off by default.

Built this way on purpose

These are not missing features. They are things SoloPDF will not do.

Local-first

Your files stay on your disk. The app makes no network requests. No cloud, no sync account, no “upload to continue”.

MIT open source

Every line is auditable on GitHub. Build it yourself and verify it really does not phone home.

Zero telemetry

No usage data, no crash reporting, update check off by default. Nothing buried three menus deep.

Notes are plain text

Annotations are Markdown, not a proprietary database. Stop using SoloPDF and your notes still read fine.

Free forever

No paid tier, no subscription, no watermark on “pro” features. And no ad revenue either.

Small and fast

A 157 MB, 1048-page PDF cold-starts in about a second. Streaming load plus virtual scrolling keeps memory flat.

Compared to the usual PDF readers

On Windows, PDF readers either show ads or lock the basics behind a paywall.

CapabilitySoloPDFAdobe Acrobat Reader福昕 / FoxitPreview / Edge
No ads or upsell pop-upsYesNoNoYes
PriceFreeFree tier + subscriptionFree tier + subscriptionFree
Highlights exported as Markdown sidecarYesNoNoNo
OCROn-deviceCloud (upload required)Cloud (upload required)No
Annotations (highlight/underline/strike/note/snapshot)YesYesYesPartial
AcroForm fillingYesYesYesPartial
Document tools (merge/split/compress/encrypt/sign)YesPaidPaidNo
FormatsPDF / EPUB / MOBI / AZW3 / CBZ / CBR / DjVu / FB2 / TIFF / TXTPDF onlyPDF onlyPDF only
No telemetryYesNoNoNo
Open source, auditableYesNoNoNo
CLI + MCP automationYesNoNoNo

Based on the free tier of each product as of September 2026. Some Foxit and Adobe capabilities require a subscription; Preview and Edge refer to the built-in OS readers.

Download SoloPDF

v0.8.0 · free and open source · MIT · no account needed

🍎

macOS

Universal · Apple Silicon + Intel · notarized
Download .dmg
🪟

Windows

x64 · Windows 10 / 11
Download .exe— MSI installer
🪟

Windows (ARM64)

ARM64 · Snapdragon X · Copilot+ PCs
Download .exe— MSI installer
🍓

Linux (ARM64)

aarch64 · Raspberry Pi 4/5 · ARM servers
Download .AppImage— deb package (Debian / Ubuntu)

macOS: drag SoloPDF.app into Applications. Notarized by Apple, so no “unidentified developer” warning.

Windows: the first launch may show “Windows protected your PC” → More info → Run anyway. That is signing reputation still building for a fresh release; it happens once.

Linux: chmod +x SoloPDF-*.AppImage then run it, or install the deb with sudo dpkg -i.

iOS / Android: iOS ships on the App Store. The signed Android APK is on the GitHub release — enable installing from unknown sources.

CLI + MCP in the box

# CLI (same rendering engine as the app)
solopdf info file.pdf              # pages / outline / metadata
solopdf extract-text file.pdf      # extract text
solopdf export-md file.pdf         # export full text as Markdown
solopdf form-fields file.pdf       # form fields & values
solopdf export-annotations file.pdf # notes → JSON

# MCP (let Claude & other AIs read your PDFs and notes)
claude mcp add solopdf -- node dev-mcp/src/index.mjs

Frequently asked questions

Is SoloPDF really free? Is there a paid tier?
Really free, and there is no paid tier. SoloPDF is MIT-licensed open source at GitHub. No subscription, no “pro” edition, no watermarked features, and no ad revenue.
Where are my annotations stored? Can I read them without SoloPDF?
In <document>.annotations.md next to the PDF — an ordinary Markdown text file, not a proprietary database. Any editor opens it and git can version it. If you stop using SoloPDF the notes still read perfectly. Snapshot annotations live in <document>.annotations.assets/ and are referenced with standard Markdown image syntax.
Does it upload my documents? Does it work offline?
No, and yes. The app makes no network requests at all, and OCR runs on-device — system Vision on macOS/iOS, a bundled PP-OCRv4 ONNX model on Windows/Linux. The update check is off by default.
Is there an x64 build for Windows?
Yes. Since v0.6.0 Windows ships both x64 and ARM64 installers (NSIS .exe and .msi for each), and Linux ships x64 and ARM64 too. macOS is a single universal build that runs on both Intel and Apple Silicon.
Which file formats are supported?
Beyond PDF: EPUB / MOBI / AZW3 (KF8) with reflowed text, CBZ / CBR comics including right-to-left manga paging, DjVu scans, FB2 e-books, multi-page TIFF, and a TXT book mode. Highlights become Markdown sidecar notes in every format.
How good is the OCR? Does it need an account or a network?
Neither. Scans convert to searchable PDF or Markdown in one step; photo capture gets perspective correction and deskewing, and two-column documents export in reading order. Separate recognition models ship for Chinese, Japanese and Korean.
Can it edit PDFs? Will it damage the original?
Document tools cover page rotate / delete / extract / reorder, merge, split, PDF ↔ images, compression, handwritten signatures and date stamps, and setting or removing the open password. Every operation writes a new file; the source is always read-only — a hard design constraint, not a default you can flip.
Can it open encrypted PDFs? Is my password stored?
It opens them, and the password is remembered only for the session. Annotations also have a privacy mode that skips storing quoted source text, which helps with sensitive documents. Removing a password requires knowing it — SoloPDF does not crack passwords.
How well does it handle Chinese?
Search applies NFKC normalization, including the Kangxi-radical codepoint trap that Chrome's PDF export produces. The interface ships in Simplified Chinese, Traditional Chinese, English and Japanese, and a 120k-entry CC-CEDICT offline dictionary is built in for look-up on selection.
Is there a command-line tool? Can an AI drive it?
Both, and they are shipped artifacts rather than internal tooling. The CLI shares the app's rendering engine and can extract text, export Markdown, read form fields, export annotations and run OCR. The MCP server exposes read and write operations as tools — read-only by default, writes behind an explicit --allow-write flag.
Does dark mode turn scanned pages into negatives?
No. Dark mode inverts selectively: text-only pages are inverted for comfort, pages containing images are left alone, so photos and figures never appear as negatives.
How does this relate to SoloMD?
Two products by the same author. SoloPDF writes highlights as Markdown sidecar files, and SoloMD is the editor that opens them — editable, full-text searchable, git-versioned. The solopdf:// links inside the notes jump straight back to the exact spot in the PDF. Neither one requires the other.
What are the system requirements?
macOS 10.15 or later (universal, Intel and Apple Silicon); Windows 10 / 11 (x64 or ARM64); a current mainstream Linux distribution (x64 or aarch64, needs webkit2gtk 4.1); iOS 15 or later.

Support & Feedback

Questions or issues? Email [email protected] or open a GitHub issue — we usually reply within 48 hours. Sample document: sample.pdf