Changelog

11 August 2026

The site went live on tasbih.info, and the first wet-hand reading came back.

• Deployed. 106 pages, 9 languages, served from Cloudflare's asset layer so a normal visit never wakes the Worker.
• The homepage had 34 words under the counter while every localised home page carried 494 to 779. It now carries 936, which matters because it is the page that has to rank for the head term.
• The root hreflang cluster declared 2 languages while 9 pointed back at it. It had been typed out when Turkish was the only other language. Generated from the locale list now, so it cannot fall behind again.
• First wet-hand measurement, on a Redmi Note 12 Pro Plus 5G: every contact reported 0 x 0, because that browser does not expose contact size at all. The size threshold is inert on this device and could be any number. The multi-touch rule did fire and correctly rejected a 2-point and a 3-point contact, so on a phone like this the guard rests entirely on multi-touch and the debounce. The threshold stays at 70 because there is still nothing measured to set it to.
• Cloudflare was injecting a managed robots.txt that blocked every AI crawler and set ai-train=no, which contradicted the open dataset this site publishes under CC BY 4.0 and asks to be cited. Turned off, so robots.txt is now the one this site generates.

10 August 2026

Content, mostly. The guide library reached the size the plan set out for it, and the pages that had metadata but no body got one.

• The 99 names of Allah, built around the chain rather than the list. Bukhari 6410 and Muslim 2677 carry the hadith and neither carries a list; the list is Tirmidhi 3507, graded da'if, and Tirmidhi says in the same book that he knows no sound chain that names them. Ibn Majah 3861 has a second list, also da'if, and comparing the two Arabic texts here gives 77 shared names with roughly 21 and 22 appearing in only one.
• Dhikr during Hajj, and 12 short guides expanded. Every hadith number in them was checked at the source. 9 claims were confirmed and filled in, 4 were wrong and fixed, and everything that could not be verified was rewritten to stop asserting a citation rather than being given one.
• Bodies for the Arabic and Indonesian pages. Both had a title and a description and about 36 words underneath. Both now carry 8 sections and 7 questions written for how those languages search, with FAQ schema in the page's own language. No dhikr, du'a, hadith or ruling appears in either, because at that point neither locale's religious content had been reviewed.
• Bodies for /tasbih-beads and /digital-tasbih-counter, which were the counter and nothing else while every other counter page carried 400 to 1,000 words.
• The 5 fiqh pages that carried no named source now state a position under the name of the scholar who reviews this site, rather than in the site's anonymous voice.
• The sitemap generates guide URLs from the registry. It had been a hand-written list of 7 while 52 guides existed, so 45 pages were shipping with no entry.
• The localised meta descriptions advertised 16 presets while the counter shipped 17. The number is read from the data now.
• The prose renderer handles italic as well as bold and links. 31 spans across 10 guides had been rendering as literal asterisks, nearly all of them transliterations.

7 August 2026

The content build. 14 new pages, taking the site from the counter and its trust pages to the full first map.

• 7 guides: what a tasbih counter is, dhikr vs zikr vs adhkar, dhikr after salah, whether a digital tasbih is allowed, 15 claims that are wrong, how to use a counter on your phone, and the Laylatul Qadr dua. Each was drafted from the corrected content and then fact-checked by a second pass against the same sources.
• 6 counter pages: istighfar, the guided post-salah sequence, adhkar, 33, 100 and Ramadan. Each opens on the right preset and target because the counter now takes both from the route.
• A /guide hub, so the guides link to each other and to the trust pages rather than sitting as orphans.
• New preset: the Laylatul Qadr dua, in the established wording. The word kareem that appears on most posters is not in it, and the preset note records why, citing IslamQA 260806 and Ibn 'Uthaymin.
• A guard so a keyword route never clears a count already in progress. Opening a page should not be a destructive action.
• The embed widget. Any masjid, school or personal site can paste one line of HTML and get an ad-free counter with the dhikr, the meaning and the hadith reference. A builder page at /embed does the picking and copying. The widget is crawlable so search engines can read its noindex directive, and it carries a small attribution link, which is the only thing asked in return.

6 August 2026

Accessibility fixes, all of them items this site had already published as open on its accessibility page.

• Both dialogs trap focus, close on Escape, and return focus to the control that opened them.
• Repeating the same announcement now speaks, using two live regions written alternately.
• The post-salah sequence position is now text as well as dots.
• Controls hidden by focus mode left the tab order instead of staying invisible but focusable.
• Reduced motion now reaches the ring sweep, the bead strand and the count animation.
• The light-theme ring fill moved to the darker accent, which clears the 3 to 1 minimum for a graphic.
• The top strip clears the status bar and Dynamic Island on an installed iPhone app.
• Holding a key on the counting button no longer runs the count up, on the button's own activation path as well as the page handler.
• New page: an offline tasbih counter explaining how the offline copy works and what it does not cover.
• Wet-hand tolerance. Wudu happens right before post-salah tasbih, and a wet finger produces contacts nobody made. Contacts far larger than a fingertip are ignored, a second simultaneous touch point is ignored, and the debounce widens after an anomalous contact. The size threshold is deliberately permissive until it has been measured on real devices.

What goes here, and what goes on corrections

This page records what changed in the app and on the site, with the date the change went live. Features, fixes, new pages, new languages.

Factual errors in the Islamic content don't go here. A wrong hadith number, a wrong grading, a missing shadda, a count shown as narrated when no narration gives it: those go on the corrections page, which is a separate log kept for a separate reason.

Keeping them apart matters. If they shared a page, a busy month of bug fixes would bury a correction, and the correction is the entry that costs something to publish.

• Here: counter features, layouts, settings, languages, new pages, bug fixes.
• On the corrections page: wrong references, wrong gradings, errors in the Arabic, mistranslations, and counts presented as narrated when they aren't.
• Both are dated, both stay up, and nothing gets edited away quietly.

5 August 2026, first release

Everything below is in the app today. Where something has a limit, the limit is written next to it.

• Counter with 4 styles: a progress ring, a classic big-number screen, a bead strand, and a handheld digital device. All 4 are skins over one counting engine, and in every one of them the whole screen counts.
• 21 dhikr presets. Each carries its Arabic, transliteration, English meaning, source reference and a direct link to sunnah.com or Quran.com. The counting screen shows the Arabic and the transliteration; tap the info button next to the phrase and the meaning, source link and count note open with it. 18 of the 21 carry a count note.
• A grading line appears only where the grade isn't simply being in Bukhari or Muslim. That's 1 preset today, La ilaha illallah, and it carries 3: Tirmidhi's hasan gharib, al-Albani's hasan, and al-Hakim's sahih with al-Dhahabi agreeing.
• 8 of those 21 carry no target, because their source gives no number. The counter reads "no set count" for them, the ring never fills, and each one carries a note. Where a familiar number is common practice rather than a narrated one, the note says so.
• Guided post-salah sequence, 9 steps, with auto-advance. The next phrase loads the moment a target lands, so a tap that arrives mid-change counts toward the new phrase instead of being swallowed.
• The live count is written to your browser's storage on every single tap. A crash, a swipe away, an accidental refresh or a dead battery should cost you nothing.
• Two-tap reset. The first tap arms it and the button relabels, the second within 5 seconds commits. Undo takes back one count, from the top strip or the bottom-left button.
• Session history and a 7-day totals chart, held in your browser's IndexedDB. It never leaves the device, and you can wipe it from the history page. A session is recorded when a count is retired: a reset, a change of dhikr, a step of the sequence, or starting the sequence with a count already running.
• Works offline after the first load, and installs to the home screen as an app. A new version activates on your next visit rather than mid-session, so a deploy can't reload the page under a live count.
• 10 languages: English plus 9 localized versions, including Arabic with a right-to-left layout. The interface and supporting content are localized; the shared dhikr names, meanings and count notes remain in English where no reviewed translation is available.
• 3 themes: light, dark, and a true-black night mode for OLED screens.
• Vibration: a short tick on every count, a double tick at every 33, a softer one on undo, and a longer settling pattern when a target lands. Two switches control it, one covering counts and the 33 tick and one covering the target, and both grey out on devices where the browser can't vibrate at all.
• A count sound, built by your browser as you tap rather than downloaded, so it works offline too. It's off by default, for the masjid.
• Space, Enter, Page Down and the right arrow count on a computer. Backspace, Page Up and the left arrow undo. Those are the keys a Bluetooth page-turner pedal sends, so a pedal counts with no setup.
• Fullscreen, where the browser has the API. iPhone Safari doesn't, so the button doesn't appear there, and it also stays hidden once the app is installed, because a home-screen app already runs without browser chrome.
• The screen is kept awake while you count, and the lock is taken again when you come back to the tab.
• The style buttons get out of the way once counting starts and come back for 10 seconds when a target lands. A focus button hides the rest of the controls.
• Open dhikr dataset at /dhikr.json under CC BY 4.0, generated from the same file the counter reads, so the published data and the running app can't disagree.
• No ads, no account, no email capture, no analytics and no third-party scripts. The whole app has 3 runtime dependencies: Next.js, React and React DOM.

What's not in it, same date

These are gaps in the same release, written down here so you don't have to find them yourself.

• No sync between devices. Your count lives in the browser you counted in, and switching phones means starting over.
• No account, no login, no password and nothing to sign up for. There's also nothing to recover if your browser storage gets cleared.
• No paid tier, and nothing to buy. The site earns nothing today. A paid tier may come later, and if it does the counter itself stays free of ads, because ads during dhikr are the thing this site was built against. Ads and a paid tier are alternatives to each other, not a pair.
• The three Quls now close the post-salah sequence. Ayat al-Kursi is available as a verified preset but remains outside the guided sequence because its post-salah narration is not in a collection this site can link to directly.
• No export. There's no file you can download of your history and no backup anywhere.
• No way to recover a reset once you've confirmed it, and no custom target: you get the narrated count, or no count.
• Two rough edges worth naming. Finish all 9 steps of the sequence and reload, and the app can still think you're mid-sequence until you pick a different dhikr. And switching off "Keep the screen awake" doesn't release a lock already held, so it takes effect the next time you come back to the tab.
• No institution has reviewed or endorsed this site. Mufti Salim Qasmi builds it and reviews the Islamic content on it, and that qualification belongs to him rather than to the website. Darul Uloom Deoband hasn't reviewed it.
• The site doesn't claim WCAG 2.2 AA. There's a written accessibility review of the interface in the project, several of its items are still open, and the claim waits until they're closed.

What's next

No dates on any of these. They're planned rather than promised, and anything missing from this list isn't planned at all.

• Evaluate adding Ayat al-Kursi to the guided post-salah sequence when its narration can be linked from a source collection this site supports directly.
• Native-speaker review of religious strings across all 9 non-English locales, so shared dhikr names and meanings can be localized safely.
• An accessibility page naming the standard the site aims at and listing the gaps it hasn't closed.
• A methodology page setting out how gradings are reported and how contested rulings are handled.
• Sync between your own devices with a pairing code, no email address and no account. None of it is built.
• Recovering a reset for a short window after it happens.

Why this log is short

The site is new, so there's one entry. That's the reason for dating releases rather than nudging a "last updated" line when nothing actually changed.

A log like this only earns anything with age. One entry tells you the site launched. 3 years of entries would tell you whether anyone's still looking after it.

So this is the honest version of a freshness signal, and right now it isn't impressive. Both of those are true, and only the second one fixes itself with time.