Random Bible Chapter

A tiny static web app that gives you a random, never-repeated Bible chapter to read, and keeps track of what you've already read so you can work through the whole Bible in random order without duplicates.

• No build step, no backend — plain HTML/CSS/JS.
• Reading progress is stored in your browser's localStorage (per browser, not synced across devices).
• A level, rank, and XP system turns your reading history into a number you can actually be proud of — see "Levels, XP, and anti-farming" below.
• Optionally, pasting a GitHub token under "Reading log sync" appends each chapter you mark as read to reading-log.json in a private GitHub repo (via the GitHub Contents API), so other tools (like a commentary pipeline) can look up what you read today. Using GitHub's API instead of a bespoke backend keeps the log reachable from sandboxed environments that only allow a small domain allowlist. The token is never stored in this repo — only in your browser's localStorage — and should be a fine-grained personal access token scoped to just that one repo's Contents permission.
• Randomness comes from the Web Crypto API (crypto.getRandomValues) with rejection sampling, rather than Math.random(). See "How random is it?" below.

Scripture memory

Highlight any words in the reader and tap "Memorize" (or tap a verse number and choose "Memorize verse", or add a passage by reference) to save it to the "Scripture memory" section. Practice runs in the app with no LLM: fill in the blanks, then first letters, then typing it from memory (passing at 90% of the words). Passed items come back after 1, 3, 7, 14, 30 and 60 days; a miss starts over at the easiest step. Passing a review that is due earns XP (2 to 10 by step, more for longer passages, 20 a day at most; early practice and retries earn none), added to your lifetime XP like trivia XP. Items live in localStorage (rbc:memory) and are included in the cloud backup. Code is in memory.js.

Running locally

Just open index.html in a browser, or serve the folder with any static file server, e.g.:

python3 -m http.server

Deploying to GitHub Pages

This repo includes a GitHub Actions workflow (.github/workflows/deploy.yml) that deploys the site on every push to main. To enable it:

1. Go to the repository's Settings → Pages.
2. Under "Build and deployment", set Source to GitHub Actions.
3. Push to main (or re-run the workflow) — the site will be published at https://<username>.github.io/<repo>/.

Levels, XP, and anti-farming

Your level is meant to be an honest signal of how much time you've actually spent reading Scripture — something you could tell a friend and have it mean something — so it's deliberately hard to inflate:

• A reading timer. When a chapter comes up, "Mark as Read" stays disabled for a short dwell period. You can't rubber-stamp your way through the Bible with rapid clicks.
• A daily soft cap on XP. The first 15 chapters you mark in a calendar day earn full XP toward your level; chapters beyond that still count fully toward your reading progress and achievements, just at a reduced XP rate. A single binge session can't out-level weeks of steady, consistent reading.
• Lifetime leveling. Finishing every chapter of the Bible banks that cycle's XP permanently and automatically starts a fresh no-repeat cycle — your level never resets, and reading through the whole Bible again keeps raising it. Each full completion is tracked and shown on your level card, so "Level 63, two full journeys through Scripture" is a real, comparable achievement.

The one true reset is "Erase everything & start over" under Reading history, which wipes your level and completions too — for when you actually want to start from zero.

Streaks, daily goal, and nudges

To make coming back every day feel good:

• Daily streak. Consecutive days with at least one chapter read. It stays alive until a full day passes with nothing read, and pulses when today's reading is still needed to save it. Streak milestones (3, 7, 14, 30, 60, 100, 180, and 365 days) unlock a one-time achievement with bonus XP.
• Daily goal. Read 3 chapters in a day for a celebration. It's purely a celebration — no extra XP — so it can't be farmed.
• Streak freezes. Read 7 days in a row to earn a freeze (hold up to 2). If you miss a day, a freeze is spent automatically and your streak survives. A frozen day keeps the streak alive but doesn't add to it, and freezes are only spent if they cover the whole gap — a longer break loses the streak but keeps your freezes.
• "Almost there" nudges. The app points out when you're 1–3 chapters from finishing a book or a few chapters from the next milestone.
• Instant feedback. A floating "+XP" pop and a light haptic buzz (where supported) on every chapter you mark as read.

Cloud backup

If you save a sync token (see "Reading log sync & cloud backup" in the app), a full snapshot of your progress — history, achievements, lifetime level, and streak freezes — is saved to app-backup.json in your private repo after every chapter. If the browser ever wipes this site's data, paste your token again and tap Restore from cloud.

• A backup is never overwritten by a device with less progress than it, so opening the app on a freshly wiped browser can't clobber your history.
• "Erase everything & start over" also erases the cloud copy.
• The token itself is never included in the backup.

Reading the chapter here (KJV)

The chapter you draw appears below the card in the King James Version, with verse numbers, paragraph breaks, and the translators' italics (the words they supplied that aren't in the original). A second "Mark as Read" button sits under the text so you don't have to scroll back up.

The text lives in kjv/, one small JSON file per book, and only the book you need is fetched. Nothing is stored server-side. Because it's fetched, the chapter text needs a connection (the rest of the app doesn't).

Where the text comes from. The 1769 King James text is in the public domain in the US (in the UK it is covered by a Crown letters patent, which affects printing rather than reading). I took it from farskipper/kjv (public domain), after cross-checking it against two other public digitizations: all three agree on the verse structure (66 books, 1,189 chapters, 31,102 verses). It matched one of them on 31,041 of 31,102 verses, and the 61 differences were spelling variants (e.g. "honour" vs "honor") or errors in the other file, not different wording. A third file had encoding damage and stray margin notes, so it wasn't used. If you spot a typo, it can be fixed in the matching kjv/<book>.json.

Questions for an LLM

Under the text, My questions is a notebook for questions you have while reading, in any Bible version (KJV, NKJV, ESV, NIV, NLT, NASB, CSB, NRSV, AMP, MSG, or type your own). Each question records its passage (defaults to the chapter you're on, or tap a verse number to target that verse) and the version you were reading.

When you're done, Copy all as an LLM prompt puts every question on the clipboard with short instructions for the model (answer from the passage first, explain what the Hebrew/Greek behind a translation says, flag where scholars disagree, don't invent sources). Tick "Also include the KJV text of each passage" to append the King James text of each passage you referenced (John 3:16, Romans 8:1-4, Psalm 23, ...). Passages it can't read are listed rather than silently dropped. Questions are part of your backup and cloud backup, and "Erase everything" removes them too.

Bible Trivia

A trivia section built to teach you things you don't already know — obscure ancient-Near-East background, archaeology, rare Hebrew/Greek/Aramaic words, hapax legomena. Every question is a mini-lesson: the explanation always shows, whether you got it right or not.

• Adaptive difficulty. Facts come in five tiers, from Warm-up to Master of Obscurity. Three new facts right in a row moves you up a tier; two misses in a row moves you down.
• Spaced repetition. A fact you get right comes back after 1, 3, 7, 14, 30, 60, 120, 180, then 365 days. Forget one and it drops back toward the start so it sticks. Facts missed in a session get one more try at the end.
• XP, with Scripture on top. A new fact earns 2–6 XP (a chapter earns 10). The big reward is remembering: a correct review earns a retention bonus that grows with the time since you last saw the fact — about 18 XP after a week, 30 after a month, and the 45 XP maximum at around six months.
• No daily limit. Learn as many new facts as you like in a sitting; tap "Finish for now" to stop. Facts you miss come back a few questions later in the same session.
• Not farmable. A fact pays its base XP only the first time you ever answer it, so total new-fact XP is capped by the size of the bank (about about 840 XP for the first 195 facts before scaling, versus 10 XP for every chapter you read). Reviews are offered only when they're due, and retries earn nothing. Once the bank passes 200 facts, every trivia payout is scaled by sqrt(200 / bank size), so growing the bank can't turn trivia into the main source of XP. TRIVIA_COVERAGE.md has the plan for growing it.

The fact bank is split by theme: trivia.js (core), trivia-prophecy.js (oracles set beside the extrabiblical records and excavations that bear on them), trivia-archaeology.js, trivia-language.js (rare words, loanwords, manuscripts), trivia-ane.js (Near Eastern background), trivia-discoveries.js (major finds and how they were made, including celebrated claims that did not hold up), trivia-greek-aramaic-texts.js (Greek and Aramaic vocabulary, manuscripts and textual history), trivia-warmups.js (easier tier 1–2 facts across the other themes) and trivia-deep.js (tier 4–5 specialist facts). Each card can carry an src field naming the inscription, text, or excavation behind it, shown as "Evidence" after you answer.

Prophecy cards are written to separate what the sources document from what scholars dispute — for example when Isaiah 40–55 or Daniel was composed, or whether Tyre's later history matches Ezekiel 26 literally — rather than claiming more than the evidence shows.

To add facts, push cards onto TRIVIA_CARDS from any trivia-*.js file and load it in index.html (never rename or reuse an id, since progress is stored by id). Keep the three wrong answers close in length and style to the right one so length isn't a clue. Trivia progress is included in backups, cloud backup, and reset.

How random is it?

The app uses crypto.getRandomValues(), the browser's cryptographically secure pseudorandom number generator (CSPRNG), instead of Math.random(). It's seeded from the operating system's own entropy sources and is good enough for security-sensitive uses like generating tokens — well beyond what's needed here, but it also avoids the subtle statistical biases Math.random() can have. The selection is also done via rejection sampling over the exact number of unread chapters, so every remaining chapter has an equal chance of being picked (no modulo bias).

True hardware randomness (e.g. from atmospheric noise or radioactive decay) isn't available to a browser sandbox, so this is the strongest randomness a client-side web app can offer.