Point Shelfmark at your folders and it builds you a proper library: it reads the metadata out of every ebook and comic, fetches cover art and page counts for anything missing, reads your EPUBs in a reader of its own, and tracks what you've finished.
Everything stays on your machine. A single SQLite file and a folder of cached covers. There's no account, no sync, and no telemetry.
Dark and light modes included. Every view adapts.
| Library — grid view with covers, search, and filters | |
 |
|
| Reader — paginated two page spread | Comics — series grouped on the shelf |
 |
 |
| Dashboard — reading at a glance | Book details — metadata, progress, rating, Open Library lookup |
 |
 |
Scans your folders, understands the files
Walks a directory (subfolders included) and indexes every .epub, .pdf,
.mobi, .azw, and .azw3, reading title, author, publisher, subjects, ISBN,
description and page count out of the file itself. Comics get a folder of their
own, because a comic is often just a PDF and only you know which is which.
Keeps comics in order
.cbz and .cbr sit in a separate Comics section, tagged from ComicInfo.xml
where a file carries one. Volumes of the same run collapse into a single shelf
tile that opens to reveal the issues underneath, so a hundred volumes of one
title don't bury everything else. Series names are read from the file's tags or,
failing that, off the filename: Vol. 3, #12, v03 and 第100巻 all parse,
and you can always set the series by hand.
Finds covers, even when the file has none EPUB and MOBI covers are lifted straight from the file. PDFs don't carry one, so the first page is rendered into a cover image instead.
Fills the gaps online Anything still missing can be looked up on Open Library, cover, page count, subjects, description, from a list of candidate matches. No API key needed. If nothing matches, type the details in yourself.
Reads your EPUBs A built in reader opens each book in its own window: paginated two page spread, table of contents, and your position remembered to the paragraph. Themes for paper, sepia and night, with adjustable size, spacing and margins, and a Publisher's font option that leaves the book's own typeface and embedded fonts exactly as designed.
Search, highlight and bookmark Full text search across the whole book jumps you to the passage and flags it. Select any text to highlight it in one of four colours, or drop a bookmark; both are listed in a side panel that takes you back. Highlights are anchored to the text itself, so changing the type size or resizing the window never moves them.
Or hands them to whatever you already use PDFs and comics go straight to your system's default application, YACReader, SumatraPDF, Preview, whatever you have, and EPUBs can too, from the details panel. An external reader can't report your position back, so those you mark off yourself.
Tracks your reading A sortable table or a cover grid of everything you own, with reading status, progress, a 1 to 5 star rating, and free text categories with a preset picker. Reading an EPUB in the built in reader updates progress on its own; nothing to type in.
Shows you the numbers A dashboard counting books read, comics read, pages read and what you're part way through, plus a breakdown by category. Totals move when you finish something, so they mean what they say.
Grab the installer for your platform from the latest release.
| Platform | File |
|---|---|
| Windows | .msi, or .exe (NSIS installer) |
| macOS | .dmg (universal, Apple Silicon and Intel) |
| Linux | .deb, .rpm, or .AppImage |
The builds aren't code signed, so the first launch needs a nudge: on macOS
right click the app and choose Open; on Windows click More info → Run
anyway; the AppImage needs chmod +x before it will run.
Then open Settings, choose the folder your books live in, and a second one for comics if you keep them, and hit Scan.
What can be read from a file varies a lot by format, so Shelfmark takes what it can get and lets you correct the rest. Every field stays editable.
| Format | Metadata | Pages | Cover |
|---|---|---|---|
| EPUB | Full Dublin Core from the OPF | Estimated | Embedded |
| Info dictionary | Exact page count | First page, rendered | |
| MOBI / AZW | EXTH header | Estimated | Embedded when present |
| CBZ | ComicInfo.xml when present |
Exact, one per image | First page |
| CBR | Filename only | — | — |
Pages only count toward your totals once something is marked finished.
CBR is deliberately left unopened. RAR's reference decoder ships under a licence
that can't be vendored into an MIT project, so a .cbr is catalogued from its
filename and nothing more. Repack it as a .cbz and it gets the full treatment.
Reflowable text has no fixed pages, so the reader stores your position as a spine index plus a ratio through that chapter rather than a page number. It survives a font change, a resize, or a switch to a two column spread. The percentage you see is derived from character counts taken when the book opens.
On Goodreads: Goodreads retired its public API in 2020 and issues no new keys, so there's no supported way to query it. Open Library is the stand in. It's free, needs no credentials, and covers the same ground.
You'll need Node.js and the Rust toolchain with Tauri's system dependencies.
npm install
npm run tauri devPDF cover rendering needs the PDFium shared library, which is
platform specific and not checked in. Drop the matching file into src-tauri/
(CI does this automatically for releases):
| OS | File | Asset from pdfium-binaries |
|---|---|---|
| Windows | src-tauri/pdfium.dll |
pdfium-win-x64.tgz → bin/pdfium.dll |
| macOS | src-tauri/libpdfium.dylib |
pdfium-mac-univ.tgz → lib/libpdfium.dylib |
| Linux | src-tauri/libpdfium.so |
pdfium-linux-x64.tgz → lib/libpdfium.so |
Without it the app still runs; PDFs just fall back to a placeholder cover.
To build an installer for your current OS:
npm run tauri buildWant something to test against? sample_books/ has a generated EPUB and PDF.
Point Settings at it and scan. Regenerate them with
cd src-tauri && cargo run --example make_sample.
Tauri can't cross compile, so installers are built in CI on native runners for each OS. Pushing a version tag builds all three and attaches them to a draft release:
git tag v0.1.0 && git push origin v0.1.0To get installers without cutting a release, for testing on a machine you don't build on, run the workflow manually from the Actions tab. It builds all three platforms and attaches them to the run as artifacts instead.
src/ React + TypeScript UI
api.ts typed wrappers over the Tauri command surface
ui.tsx shared primitives (icons, buttons, star rating)
reader-prefs.ts themes, typefaces and spacing for the reader
series.ts groups comics into series from tags or filenames
pages/ Dashboard, Library (table/grid + drawer), Settings
Reader.tsx paginated reader; Books-style presentation
ReaderWindow.tsx host for the standalone reader window
src-tauri/src/
commands.rs Tauri IPC boundary
db.rs SQLite schema and queries
scanner.rs folder walk, registration, cover refresh
formats/ epub / pdf / mobi / comic metadata extraction
reader.rs spine, contents, sanitising, asset serving
pdfcover.rs PDF first-page rendering via PDFium
metadata.rs Open Library search and cover download
Your library lives in the platform app data directory, library.db plus a
covers/ folder. Shelfmark never modifies the files themselves.
The bundle identifier is still com.kas.bookvault, from when this was called
Book Vault. That string is what picks the app data directory, so changing it
would strand every existing library. It's invisible to users; leave it be.
cd src-tauri && cargo testMIT © Carson