Inkspell — a chapter-locked glossary for EPUBs you already own (Android)

[Franco]

Hi everyone,

I'm the developer of Inkspell, an Android companion app for long novels. It's not a reader — you keep reading in whatever you already use. Inkspell works from your EPUB file and answers "who is this again?" without telling you anything past the chapter you're on.

You upload an EPUB you own, set your chapter, and it builds:

• A glossary of every character, place, faction and object you've met so far
• Story-so-far recaps that end at your chapter
• A relationship map – allies, rivals, mentors, family
• Character arcs – how someone changes, chapter by chapter
• Sagas – link the books of a series so book four remembers book one

Everything is filtered by your current chapter. Move your bookmark forward and more appears; nothing from later chapters is ever shown.

Why I built it

Kindle's X-Ray only exists for books bought from Amazon (and not all of those – it's generated per title), and fan wikis tend to spoil the ending in the first paragraph. For a sideloaded EPUB there was nothing. Longer write-up on the X-Ray gap: https://theinkspell.com/guides/x-ray-for-any-book/

Worth knowing before you install

• EPUB only for now (no PDF). You need the actual EPUB file – books bought on Kindle don't come as one, so those won't work. No store in the app – you bring the book.
• One book per file: omnibus / box-set EPUBs are blocked at upload. If you have the volumes as separate files, upload each one and link them as a saga instead.
• Processing happens on our servers: the file is stored in Google Cloud Storage and the text is sent to Google's Gemini API to generate the entries. They're machine-written and can be wrong; every entry has a report button.
• The first chapters are usually ready in 5–10 minutes; the rest finishes in the background while you read.
• It's a subscription: $9.99/month or $89.99/year, normally with a 7-day free trial, and there's no free tier. I'd honestly prefer to make it free, but Inkspell is a tiny self-funded project and every book costs real money to process, so I can't.
• Android only for now; iOS has a waitlist on the site.

For MobileRead readers: use the code MOBILEREAD and you get 30 days free instead of 7.

How to redeem it – this part matters:

• Enter it during first-time setup: on the plan screen, tap "Have a promo code? Redeem it" and type MOBILEREAD. That's the only place the field appears – once you're past setup, there's nowhere to enter it.
• It replaces the 7-day trial rather than adding to it, and it's for new accounts only.
• Like the trial, Google Play asks for a card. Nothing is charged for 30 days, then it renews at the monthly price unless you cancel – you can cancel any time in Play.

What I'd most like feedback on

• Is it actually useful mid-book? Which parts did you reach for – characters, places, the relationship map, the recaps, character arcs – and which did you never open?
• Entries that are wrong, miss someone important, or give away something you hadn't read yet. That last one is the bug I care about most – the report button on each entry comes straight to me.
• What's missing: something you wished it could tell you and it couldn't.
• EPUBs where chapter detection goes wrong – structure varies a lot between publishers and editions.
• Omnibus editions – if that's how most of your series are stored, tell me. Splitting them into volumes is the obvious next thing to build, and I'd like to know how many people it would unblock.

📲 Google Play: https://play.google.com/store/apps/d...%3Dlaunch_post
🌐Website: https://theinkspell.com/

Bug reports very welcome – happy to answer anything.

[Quoth]

"Processing happens on our servers: the file is stored in Google Cloud Storage and the text is sent to Google's Gemini API to generate the entries."

Total fail in privacy (US Gov can know what you are reading), environment and morality as Gemini is built using violated copyright works.

Needs Internet.

Subscription.

Some epubs don't even have chapters.

[Sarmat89]

Originally Posted by Quoth

Gemini is built using violated copyright works.

Reading books does not violate any copyrights, neither does remembering what one has read.

[Franco]

Originally Posted by Quoth

"Processing happens on our servers: the file is stored in Google Cloud Storage and the text is sent to Google's Gemini API to generate the entries."

Total fail in privacy (US Gov can know what you are reading), environment and morality as Gemini is built using violated copyright works.

Needs Internet.

Subscription.

Some epubs don't even have chapters.

Fair points, let me explain:

Privacy: It isn’t a local tool. Your EPUB is uploaded and kept privately in Google Cloud Storage, so the entries can be built and maintained. It’s used only for that, it’s never shown to or downloadable by anyone, and it’s deleted when you delete the book or your account. Its text is processed with Google’s Gemini API, and what’s sent there carries no account identifier: Gemini isn’t told who uploaded the book or who is reading it. No platform ever receives your book titles or reading activity. Still, anything held with a cloud provider is subject to whatever legal process that provider is, so if that’s a dealbreaker, a fully local tool is the right choice and this isn’t one.

Gemini and copyright: I can’t dispute that and won’t try to talk you out of it. I use Google’s model, I don’t train my own, and I can’t vouch for what went into theirs during training. It’s a legitimate reason to pass.

Internet: Building a book needs a connection, because that part runs on a server. Once a book has synced, the app works fully offline: the glossary, recaps, relationships and arcs all live on your phone. It only needs to reconnect once in a while to confirm the subscription.

Subscription: Yes, it is one. I explained why in the post, but I get that it’s a no for some people. I wished I could offer it for free and I'm actively working to improve performance so I can reduce its price.

EPUBs without chapters: Also true, and it’s a real limit. Everything is filtered by chapter, so a file with no chapter structure has nothing to filter by. The app tries to rebuild chapters from the headings. If it isn’t confident, it stops and shows you what it found so you can mark where chapters begin, or cancel and get the slot back. A file with no structure at all is a bad fit.

I’m trying to build something that lets you keep enjoying a long book without stopping to look things up or getting spoiled by what you find. The reading itself stays yours, in whatever app you already use.