What changed

One block per release, newest first, in plain words. Your palace updates from the signed release once you say yes on its card; Settings shows which release it is on. This file is the source of truth: the public page at https://www.usemagicbox.com/changelog is this file, read from the kernel's release branch. A release cannot be signed without its block here (bin/release.mjs checks). The nerd detail behind each line is in docs/BUGS.md, by row number.

Release 52 · 13 Nov 2026

• The guide for assistants says where things are now. The manual an assistant reads (For assistants on the website) now names the five tabs and Connections → Your assistants and Your data, says Ask your palace instead of the chat tab, and no longer tells an assistant the palace builds pages on its own. Words only. (BUGS 380)
• Your palace stays light. It no longer turns dark when your phone is in dark mode: one sand-and-sage look, day and night. (BUGS 381)
• Plans, and what your palace's thinking costs. Your palace now has a plan, Mini or Standard, with a free week to start. It tells the maker one number a month, what its thinking cost (0 if you think on your own account), with the daily counts it already sends, and never what you asked or what it said. Settings → About says so. (BUGS 364)
• Settings → Plan and billing. A new section shows your plan and what it includes, the days left in your free trial, this month's thinking against your plan with the day it comes back, the file space you use, a Buy more thinking button, and a link to the website for your card, receipts and plan. Settings → About's status says your plan too. (BUGS 369)
• A clear stop when the month's thinking is used. At 80% Home says so once. When it is used up, chat says "I've used this month's thinking", Home offers to add more, and both say the day it comes back. Routines and the night's tidy keep running on the small model, and nothing costs more without you saying so. If you think on your own account, there is no cap. (BUGS 365, 372)
• File space follows your plan. 2 GB to start and 10 GB more with each space add-on; your palace restarts once when it's added. When it is full, the message says how big it is. (BUGS 366, 372)
• A palace can be frozen by its maker. If a palace isn't paid for, or the maker has to stop it, it is frozen: nothing runs, and your memory stays exactly as it is. Assistants are told the palace is frozen and where its owner can open it again; routines, the nightly tidy and reminders wait; the chat, WhatsApp and Telegram answer with one line that says why and how to open it again; Home says so with a Pay button when a payment didn't go through, or a Contact button when we froze it, and both open the website. A hiccup on the website never opens a frozen palace by mistake: only a clear "open" does. Your own Freeze in Settings → Safety is separate and works as before, and Home now says "Your palace is frozen. Unfreeze it in Settings → Safety." instead of talking about safe mode. (BUGS 367, 368)
• Notes from the maker on Home. A short note from the people who make your palace, like an offer to download everything, shows on Home for a week. When a note stops being true, because you paid or undid closing your account, a new note says so. (BUGS 367, 372)

Release 51 · 12 Nov 2026

• Chat, To-dos and Pages left the menu. Your palace now has five tabs: Home, Memory, Connections, Routines and Settings. Connections is where your assistants and your accounts live, and Ask your palace (the chat) sits at the bottom of it, as one conversation. Your to-dos show up in the Sunday routine, and the pages you already have are under Settings → Things I built. If you had pages or to-dos, Home says where they went, once. (BUGS 304, 305)
• Your palace stops offering to build pages. It still makes one if you clearly ask, and your existing pages keep working. (BUGS 306)
• Routine results on Home and in Routines. Home shows what your last routine said and which one runs next, and each routine in Routines keeps its last result, whether it ran on time or you tapped Run now. Results also go to Ask your palace, and to WhatsApp or Telegram when they are linked. (BUGS 307)
• One connection per assistant. When you connect Claude, ChatGPT, Grok or Muse again, the new connection replaces the old one, so your list doesn't fill up with copies that nobody uses. Your diary says which one was replaced. Picking your palace on usemagicbox.com no longer shows the demo palace, and if you have one palace it is picked for you. (BUGS 310)
• See what each assistant really does. Each assistant under Connections now says when you connected it, when it last read your memory and when it last saved something, and whether it saves at all ("Saves memories: yes, 3 today" or "Reads but never saves"). (BUGS 311, 312)
• A quiet assistant gets a nudge. If an assistant that used your memory hasn't for 5 days, or keeps reading but has never saved anything, Home says so once, with the words to paste into it again or a way to disconnect it. It won't nag: once every two weeks at most. (BUGS 312)
• A palace several releases behind is reminded once a week. If your palace is two or more releases behind and you update by saying yes, Home says so and puts the update card back once a week. (BUGS 313)
• Better answers from your memory. When an assistant searches your memory, the note that is about what it asked for now comes first, your own words come before what other assistants saved, and each note comes with a line saying what it is about, so your assistant gets the picture in one look. (BUGS 317)
• Your assistants are told to save as they go. What you paste into each assistant now asks it to look at your memory once at the start, search it for each question, save what you tell it before it replies, and at the end of every conversation save anything new it learned about you without being asked. It also says to set up anything you want on a schedule as a routine. (BUGS 318)
• See your assistants at work. Home now opens with how many things your assistants can see about you, with a way to add more, and how often they read your memory and saved something today ("Today Claude read your memory 12 times and saved 3 things."). (BUGS 319)
• What your palace tells the maker, now with a little more. Your palace's daily report to the maker now also counts which accounts are plugged in and how often they are used (Google, calendar, WhatsApp, Telegram, iMessage), when Google says no to connecting, files kept, memories brought in and how many notes they made, the getting-started steps and the tour you did or skipped, nudges shown and acted on, and routines switched on from a starter or made from your own words. Numbers only, as before, never what any of them say; Settings → About says so. (BUGS 322, 323)
• Bring in what your other AIs know. There is a short message to paste into Claude, ChatGPT, Grok or Muse and send once: it asks your assistant to tell your palace everything it remembers about you, one thing at a time, and to say how many it saved. Your palace files them into the right rooms within minutes, and can show how many came in and where they went. (BUGS 314, 316)
• The same thing twice is kept once. If an assistant saves something it already saved the same day and your palace hasn't filed yet, your palace says "Already noted." and keeps one copy, so pasting the message twice doesn't double your memory. (BUGS 315)
• A connection that saves faster than 600 things in ten minutes is asked to slow down. Your assistant is told that the thing it just sent wasn't saved, and to wait a few minutes, send it again and carry on. Reading your memory is never slowed. (BUGS 315)
• Your palace walks you through the start. A new palace opens on one screen: connect your first assistant, about five minutes. It then walks you through connecting it, seeing it work ("Claude is connected and just read your memory."), bringing your memories in, connecting your email, calendar and chats, and picking a routine. Steps you already did are ticked, any step can be skipped, and See how it works shows what the palace does in four steps. (BUGS 308, 329)
• One card per assistant. Claude, ChatGPT, Grok and Muse each have a card that says whether it is connected and whether it saves, and one Connect button that shows every step with a picture. Apps like Claude Code are under Another app. (BUGS 328)
• Home says what each assistant really did. "Connected 3 Nov · last read 2 min ago · last saved yesterday", or "hasn't read your memory yet" with a way to check it. (BUGS 309)
• No more sample page card. A new palace no longer starts with a "New page: Groceries" card. (BUGS 327)
• Three routines you can switch on with one tap. Routines now opens on "Start with one of these": what's on today and anything urgent in your mail (every weekday at 7am, once your email and calendar are connected), the promises and to-dos from this week you haven't done (Sunday at 6pm), and what your assistants learned about you this week (Friday at 5pm). Tap "Yes, do this" and it runs once straight away, so you see what it does. (BUGS 320)
• Make your own routine in one line. Type what you want and when, like "every Friday at 5, tell me what my assistants learned about me", and tap Make it. Your palace reads the time from your words and puts a card on Home; say yes and it's on. If it can't tell when, it asks you to say when; for something that happens once (like "tomorrow at 3") it tells you routines repeat and to say it to Ask your palace, which sets reminders. (BUGS 330)
• Make your own understands more ways of saying it. "Weekly shop list every Saturday morning" runs on Saturdays, "tell me what's on" keeps its last word, and "every morning at 7 and every evening at 7" keeps both times. When a day sits outside the schedule ("bins night is Thursday, remind me at 7pm") or two times read the same, it asks you instead of guessing. (BUGS 332)
• An import that brought nothing in no longer ticks "Your memories came in". The first-run step stays open and says nothing came in, so you can try another file or ask your assistant instead. (BUGS 333)
• Home counts your memories the moment they arrive. When an assistant tells your palace 30 things, Home says your assistants can see 30 things about you straight away, and "Things I learned about you" shows the newest of them, instead of waiting for the night's tidy. (BUGS 334)
• Your first day is Day 1. A palace set up in the evening no longer says "Day 2 together" that same night; days are counted in your own time zone. (BUGS 337)
• Routine times read like a clock. Routines say "every weekday at 7am" and "every Friday at 5pm" instead of 07:00 and 17:00. (BUGS 321)
• A connected assistant can no longer empty your memory in one go. If an assistant asks to forget more than 50 lines at once, nothing is forgotten: a card on Home asks you first, and the lines go only if you say yes (Undo still brings them all back). Assistants that forget or change things more than 60 times in ten minutes are asked to slow down. (BUGS 338)
• Many small forgets are limited too. A connected assistant can forget at most 100 lines in any ten minutes, however it splits them up. Past that, each forget waits on a card on Home for your yes, and Undo brings back everything an assistant forgot. (BUGS 348)
• One-off reminders point to the right place. When you type a one-off like "tomorrow at 9 remind me to call the dentist" into Make your own, it tells you to say it to Ask your palace, at the bottom of Connections, which sets it. Before, it sent you to your assistants, which can't set reminders. (BUGS 350)
• "Bins night is thursday, remind me at 7pm" just works. A day you name anywhere in the sentence, with a time, makes a weekly routine on that day. Every example the box suggests now works if you type it back exactly. (BUGS 351)
• "Read your memory" means it read your memory. An assistant that only asked what's waiting for you or listed your routines no longer counts as having read your memory: the first-run check turns green, and the quiet-assistant nudge stays away, only once it really looks at your memory. (BUGS 341)
• A plainer "Let Claude use your memory?" page. When you connect Claude, ChatGPT, Grok or Muse, the page names it, says in one line what it will do, and has one choice ticked for you: "Read and save my memory". Reading your messages and building for your palace are under "More choices", off. (BUGS 342)
• Your palace's words follow the new tabs. Replies, diary lines and cards now send you to Connections → Your assistants, Connections → Your data → Gmail and Calendar and the Routines tab, never to places that are gone. Saying yes to a new routine tells you "It's on", and Ask your palace suggests questions you'd ask. (BUGS 344)
• Plainer reminders on Home. A quiet assistant's reminder says "Paste the reminder into Muse again" with a Show the reminder button, and the update reminder says a newer version with fixes is waiting, without release numbers. (BUGS 345)
• Routines in plain words. Each routine you switched on says what it does in one line instead of the instructions your palace follows, and a finished run shows the first words of what it said. (BUGS 346)
• Your data, one card each. Connections → Your data shows Gmail and Calendar, WhatsApp, Telegram and iMessage as one card each: what it's for, one Connect button, the steps on your phone, and once it's connected, when it last synced and which routines need it. (BUGS 324, 326)
• When Google says no, you're told why in plain words. If Google refuses the connection because of a setup problem on our side, your palace says so, that it isn't your fault and to try again later, instead of showing Google's error code. If Google never sends you back, the Gmail card says so a few minutes later. If you said no on Google's screen, it says so and how to try again. (BUGS 325)
• The first run shows where your assistant looked. When your first assistant reads your memory, the palace plan behind the first-run card lights the room it read (Family, Health…) and the keeper goes there. Only the room's name is kept, never what was read. (BUGS 353)
• The box has a new look. Sand and sage colours, serif headlines, ruled lists instead of boxes, and the palace plan behind each first-run step, always whole and never under the card. The fonts come from your box itself, so nothing is fetched from anywhere else. (BUGS 361)
• The page where you let an assistant in now has the box's new look. Sand and forest colours and the same serif headline as the box, so connecting Claude or ChatGPT no longer flips to the old orange colours halfway. (BUGS 362)

Release 50 · 5 Nov 2026

• What last night's tidy did, on Home. Home says in one line what your palace's tidies since last night did ("Last night I tidied 14 notes, added 3 links and retired 2 old facts."; "Since last night I've tidied …" when one of them ran this morning or later). Close it and it stays closed, on every device, until the next tidy. (BUGS 283, 295)
• A palace you can move around. Your palace now sits on a bigger stage of its own. Drag to slide it, pinch or scroll to zoom, and tap Fit to see all of it again. Room names keep a readable size however far you zoom, and each device remembers where you left it. (BUGS 296)
• Palace and Decisions. The List is now Decisions: only what needs you (questions, last night, what's waiting to be filed, Set aside and Dropped, Import and Forget). What I know about you, Look back and Who uses your memory moved to About this palace, on the palace. New and Used most are buttons beside the search. (BUGS 297)
• Imports that tell you. Import your memories now opens over Memory and you stay there. A line says it has started, and a card at the top of Memory shows how it is going, then how many notes came in and which rooms they went to. (BUGS 285, 298)
• History no longer turns up blank. Opening History under Last night could leave a blank screen once your palace had checked your day for things to learn, which it does every night. It now lists every night. Opening a note from New or Used most could do the same; it opens now. (BUGS 299)
• Undo the latest tidy. If the latest tidy filed something wrong, undo it under Memory → Decisions → Last night: the notes it changed go back as they were (merged notes split again), notes it started go, and the lines it filed come back to Waiting to be filed. Your palace keeps a way back for the latest tidy only (a tidy also runs a few minutes after you tell it something), says when that tidy ran, and says plainly why when it can't (for example, you changed one of those notes since). (BUGS 284)
• Undo an import. An import now says how many notes came in and which rooms they went to, and the latest one can be undone for a week (Undo this import, on its card or under Memory → Decisions): the notes it started go, and the lines it added to notes you already had are taken out. A note you changed after the import is kept. Changed your mind? Undo it in Settings → Things I built. This works for imports made from this release on. (BUGS 282, 285)
• Remove a note. Take a whole note out of your memory, with the links other notes had to it. Its documents and photos stay, as Loose papers. Undo it in Settings → Things I built. (BUGS 282)
• Emptying your palace stops what it was doing. Emptying your palace while it is tidying or importing now stops that work, and nothing from before comes back. (BUGS 289)
• Undo never loses a removed note. Undo after removing a note brings it back even when a new note with the same name has appeared since: the old one comes back beside it, and the answer says so. (BUGS 290)
• Removed means removed. A note you removed is gone for good once Undo can no longer bring it back, and what a connected assistant forgets can no longer push your own removal out of reach of Undo. (BUGS 293)
• Forget means forget. Forgetting something also takes it out of the tidy's history and its way back. (BUGS 294)
• Whose words, even in a correction. When a correction replaces a line another assistant or a message wrote, your palace treats that old line as someone else's words, in its own chat and in what it tells your other assistants, the same as everywhere else in your memory. When you correct a line someone else wrote, the old line in the note's history is still treated as their words. Undo keeps track of which lines someone else wrote, too. (BUGS 288, 291, 292)

Release 49 · 4 Nov 2026

• Give your palace a new name. Change it in Settings (Rename it), or ask in your palace's chat ("call yourself Juno") and say yes on the card. The new name shows everywhere in your palace at once, and your diary notes the change. (BUGS 280)
• Routines fit on a phone. A routine with recent jobs no longer squeezes its name into a thin column; its recent jobs sit underneath. (BUGS 281)

Release 48 · 3 Nov 2026

• Your assistants look before they ask. Claude, ChatGPT, Grok, Muse and your other assistants are now told, by your palace itself, to look in your memory before answering anything about you, and to save what you tell them as they go, without you asking each time. Your palace's own chat is now told to search first instead of asking you to repeat something it or another assistant was already told. (BUGS 275)
• Tell it once. When you connect Claude, ChatGPT, Grok or Muse and say yes with your passkey, your palace now shows one more thing before you go back: a short paragraph to paste into that assistant, with a Copy button and where it goes (for Claude, Settings → Your Claude → Instructions). With it, the assistant looks in your palace and saves to it on its own. The same text is always at your palace's address followed by /connect/instructions.txt. (BUGS 276)

Release 47 · 1 Nov 2026

• Your box is now Mind Palace. Your box is now your palace. Same memory, same you, a new name, a new face on Home and the same address (usemagicbox.com). Nothing to set up again: the first time you open Home after the update, one line says so; close it and it's gone. On an iPhone or iPad the home-screen icon doesn't update by itself: remove the old one and add your palace to your home screen again to see the new one. Passkeys you make from now on are labelled "Mind Palace" (not "magic box" / "Box owner"); older ones keep their old label and still work.
• Your memory, drawn as a place. Memory now opens on your palace: a room for each part of your life, the people, places and topics in each, and the threads between them. Search lights up the rooms that match; open any note to see what links to it, fix a fact in place, or add and remove a link yourself (each one gets a line in your diary and in both notes' history). The Palace / List switch at the top brings back the list you know, and remembers your choice on each device. Looking costs nothing.
• Keep documents and photos in your palace. Keep a document or photo next to the note it belongs to (up to 25 MB each, inside your 2 GB). Your palace reads the words in PDFs, Word, Excel and PowerPoint files, emails and text files so you can search them; it does that on your palace itself, never on the internet. Photos are found by their name and date; if you switch it on in Settings → Memory, your brain can describe a photo when you ask, for a few cents each. A photo or file you send in chat is kept only when you tap "Keep in my palace". Assistants you tick "My messages, mail and documents" for can read your documents too (same checkbox as before, nothing you already allowed changes); one that wants to keep a file asks with a card first, and the file is marked as from outside. Once your palace is locked, they sit on its locked part with everything else. Download my palace includes them (it now streams, however big it gets) and brings them back in whole; Empty my palace removes them.
• Mind Palace everywhere. Chat, cards, the diary, notifications, routines, pages, connector cards, the update card and Settings say "your palace" now, and the page that asks for your passkey when you connect Claude, ChatGPT, Grok or Muse says Mind Palace in the new colours. A Telegram link you make from now on shows as "Mind Palace" in Telegram's Devices list. In your own WhatsApp or Telegram self-chat you can start a message with "palace," as well as "box,". The look-change card is "Change how my palace looks".
• Small things. To empty your palace (Settings → Safety → Empty my palace) you now type "empty" and your palace's name; "erase" still works. A download you open from your palace's address is called <name>-palace.zip. The few lines that still said "the box" (frozen, safe mode, being locked, the brain line on Home, Google's unplug lines) say "your palace". The document reader, which already had no way onto the internet, now can't look up names on it either. (BUGS 255–274)

Release 46 · 29 Oct 2026

• Demo boxes say what really happened. On a demo box every card is turned down the moment it's made. Adding a routine, asking for a tidy, or proposing a page, connector or look change now says so plainly ("the card was declined automatically, so nothing was built") instead of saying a card is waiting on Home.
• Plainer tool descriptions. What your box tells Claude, ChatGPT and the others about each tool now just says what the tool does and returns, without telling the assistant how to behave. Nothing an assistant can do has changed.
• Reading your messages is never pre-ticked. When an assistant or an app on your computer connects, "My messages and mail" starts unticked. Tick it yourself if you want it to read them.
• Your other assistants can tell the maker too. Claude, ChatGPT, Grok or Muse can now suggest a bug report, a feature request or a question for the people who make your box. It's a card on Home showing exactly what would be sent, in the assistant's own words plus your box's status lines; it goes only if you say yes with your passkey.
• Clearer about missing permissions. When an assistant connects through usemagicbox.com and "My messages and mail" isn't offered, the connect page now says why and which setting turns it on (Settings → Other assistants). An assistant that hits that wall is told the same, so it can tell you.
• Assistants that look for sign-in in an unusual place now find it. Some assistants ask usemagicbox.com for the sign-in details at a second address; it now answers there too, once the site is updated.
• Clearer steps for connecting other assistants. Settings → Other assistants now gives one address for Claude, ChatGPT, Grok and Muse, https://www.usemagicbox.com/mcp, the same five steps for each, and what a connected assistant can and can never change. A new switch there, "Let assistants connected through usemagicbox.com read my messages and mail" (off by default, turning it on asks for your passkey), is what Muse needed to be offered "My messages and mail": switch it on, then disconnect Muse and connect it again.
• Correct your memory in plain words. Tell the chat "The blue lamp arrived and was sent back" and your box finds the fact that's wrong and changes it; the old line stays in that note's history. Claude, ChatGPT and the others, with "Memory" ticked, can do the same. If more than one fact could be the one, nothing changes and it asks which.
• Your box tells you when a page is behind a correction. After you correct a fact in chat, it says which page shows that note and offers to re-check it, which loads it again with today's data. Your other assistants can ask whether a page is behind too. Pages that show parcels or orders from your email are now built to let what memory says about an item win over what the mail says.
• Your box can say what needs you. One list of everything waiting on you: cards for your yes, questions, jobs that are stuck or failed, pages with questions for you, pages just finished or whose build failed, and anything urgent. Home now shows pages that need you, with a button to open each page's chat. Claude, ChatGPT and your other assistants can read the same list so they can remind you. They get only each item's one-line title, never what's inside it. A title can name who a message is for, an email's subject or the start of a fact waiting for your yes, but never the whole card, message, email or note. Assistants without "My messages and mail" see only that an email or message is waiting for your yes, not to whom or what it's about, and only that your box has a question about who someone you message is, not their number or name.
• A correction made in a chat turn that read email or messages now always waits on a card. If the chat has just read an email, a message or a file, any correction to one of your facts is a card on Home for your yes, even when you said it yourself, so nothing it read can flip a yes to a no, move a time or drop a day. When it hasn't read anything like that, your correction happens at once as before.
• A card that corrects your memory shows the whole new fact and the fact it replaces. Before, it showed only the first 120 characters of what it would write. Your yes changes only that fact: if it has changed since the card was made, nothing changes and your box tells you to say it again in chat.
• Re-checks by other assistants are named and limited. When Claude, ChatGPT or another assistant re-checks one of your pages, your diary says which one asked, not "you asked". Each can re-check pages at most 10 times an hour (30 a day for all of them together); your own Re-check has no limit.
• A photo you send no longer sticks to the bottom of chat. Your message with a photo now takes its place in the conversation like any other, and the words you typed with it stay in the bubble above the photo.
• Chat uses the whole width. On a phone your messages and the box's replies fill the screen instead of half of it; on a computer the chat is wider and photos show bigger.
• The chat box grows as you type. A long message grows the box to about six lines, then it scrolls; after you send it goes back to one line. Works in Safari and Firefox too.
• Your box says when its brain can't see photos. If the brain you picked in Settings → Brain can't look at pictures, your box says so plainly and keeps the file, instead of "My brain didn't answer". When one model can't take the photo, the next one that can answers instead. And a follow-up like "what does it say?" now shows your last photo to the brain again, so it can answer.
• Your box tells the maker how much it's used, in numbers only. Every 10 minutes, with its usual "I'm alive" ping, your box now sends usemagicbox.com its time zone (so the days line up) and daily counts: how often you opened it and sent a chat message, how many tool calls each connected assistant made (Claude, ChatGPT, Grok, Muse or other; directly or through usemagicbox.com; reads and writes), and how many routines ran, jobs finished, pages were built and cards you approved or declined. Never your memory, messages, chat, page names, card titles or what an assistant asked for. Settings → About says what is sent, and your diary gets a line the first time it goes.

Release 45 · 26 Oct 2026

• A simpler menu. Your box's menu is now Home, Memory, To-dos, Pages, Routines, Connectors and Settings. Cards waiting for your yes sit at the top of Home, so there is no Cards tab any more; yes still asks for your passkey. Chat is no longer a tab: it's a small button on every screen that opens a chat panel. Each page has its own chat, opened from its card under Pages. Notifications and old links to Cards now open Home.
• Set up tasks from anywhere. Your box is one memory for everything, and an always-on agent that learns about you and does the tasks you set up. Claude, ChatGPT, Grok or Muse (with "Memory" ticked) can now add a routine, list them, run one now, cancel one and read how it went, all sharing your box's memory. The first time each routine is added it's a card on Home that says who sent it, what it does and when; nothing runs until you say yes. A cancelled routine can be brought back in Settings → Routines. At most 20 routines, at most once an hour. Routines still only read and report: they never send, change or buy anything.
• Jobs. Every time your box runs a routine it is now a job with its own log: each step it took, what it read, what it was told no to. If your box restarts or its brain is slow mid-job, the job carries on from where it stopped. A job's result lands in your memory (a note per job in a new Jobs map, one dated line each time it runs) as well as your chat. If a job can't go on without something only you know, it asks you one question on Home and carries on with your answer. A job that fails is a card; Retry runs it again.
• Checks. A routine, a job or a card can name a check that has to pass before it counts as done, for example "the result mentions the tracking number". Checks run boxed, with no way onto the internet.
• Notifications on your phone. Turn them on in Settings → Notifications: your phone hears when a card is waiting for your yes, a page has a question or is ready, or something urgent needs you. They only ever show a short title, never what a card or message says. Pick which ones you want, and set quiet hours.
• Building a page is a conversation. Ask for a page and your box first asks you two to four quick questions, each with its own suggestion, then builds version 1. Say "just build it" to skip them. Then tell it what to change in the page's own chat ("make the header smaller", "add a filter") and each change becomes the next version. The page's questions, progress and the finished page all go to that chat from the start; the chat you asked in gets one line with a link to it. A page's card under Pages shows its chat so you can answer from there, and a Re-check button loads the page again with today's data, runs its checks, and brings back a page that was switched off after a bad first load.
• Better-looking pages. Pages are now designed to a written plan before they are built, built on the strongest brain your box has, and reviewed for how they look and feel on a phone. A page that looks poor goes back for another round before you see it. A page can now use up to $3.00 of brain spend (it was $1.00).
• Your keys stay locked away. Settings → Keys: a key for an outside service is kept sealed on your box, under a name you pick, and is only ever sent to the one site you name. Your box and the connectors it builds use it by its name; nothing your box, a page or another assistant can read ever shows it. Adding, replacing or removing a key needs your passkey.
• Memory tools. You or your assistants can change one fact in a note instead of forgetting it and saving a new one; the old wording stays in that note's timeline. Your box can say where what it knows came from (you, WhatsApp, Gmail, your other assistants…), what it learned last night and who saved what this month, and give a look back over a day, a week or a month. An assistant can show what a tidy would file, merge and archive (doing it is a "Tidy now" card for you), and put notes you name under a map of their own ("put these under the house hunt").
• The diary shows what I wouldn't do. When I say no to something (a card you turned down, a switch set to Never, something another assistant or a page wasn't allowed, a connector kept from a website), the diary has one line saying what and why.
• Done means done. A to-do is ticked off only when your box can see it happened, or you say so. Something nobody could check stays on your list marked "said, not verified" until you tick it.
• Replies stay where you asked. If you switch to another chat while your box is still thinking, its answer still lands in the chat you wrote in.
• One address for every assistant. Claude, ChatGPT, Grok and Muse can now connect to your box through one address, https://www.usemagicbox.com/mcp: it asks which box is yours and sends you to your box to say yes with your passkey, same as connecting directly. Reading your messages and mail through it stays off unless you turn it on. Directory listings in Claude, ChatGPT and Muse come later.
• Clearer for assistants. Every tool your box offers now has a short title and says plainly whether it only reads, adds, or changes what's there. Claude and ChatGPT can identify themselves by publishing a description of themselves, which your box checks safely, instead of registering a new client each time.
• A demo box for app reviewers. A box set up as a demo on its own machine holds made-up sample data, lets a reviewer connect without a passkey, and turns down every card on its own. It can't be switched on remotely, or on a box that has a passkey or real data.

Release 44 · 26 Oct 2026

• Big pages build properly. Building or fixing a big page could stop with "the brain didn't return a complete page": the page was longer than the room the brain had, so it got cut off. Pages now get twice the room, and a page that still gets cut off is written again, more compactly. If a round of fixes can't finish, your box keeps the last version it tested (when it works and passes most of its checks) instead of throwing the work away.
• Two Google accounts: you hear when one can't be read. If Gmail refused one of your accounts, its mail used to quietly go missing from searches and pages. Now the answer says which account couldn't be read and Google's reason, so you know the list is missing something.

Release 43 · 23 Oct 2026

• A new chat. What you send shows up straight away, before your box has answered. While it thinks you see "…", and its reply fills in where the dots were. Messages from the same side sit together instead of each in its own block. If you scroll up to read, a new reply no longer pulls you down; a "New message" button jumps you to the latest when you want. On iPhone the box you type in stays above the keyboard instead of hiding behind it. If a message doesn't go through, it says "Didn't send" on the message with a Retry button; your words aren't lost. Page chats work the same way.
• Grok and Muse in Settings → Brain. Use my own key now lists Grok (from xAI: make a key at console.x.ai → API Keys, it starts with xai-) and Muse (from Meta: make a key at dev.meta.ai → API keys), each with a link to where you make the key. Release 42 put them there by hand; they are now part of the box's screens properly, so they stay.
• Grok and Muse can now use your box's memory: Settings → Other assistants has the steps.
• A new home. Your box's first screen is more fun now: what I learned about you lately, why I'm safe, what's connected, anything urgent you should know about, and a few numbers about us (days together, things I remember, pages I made you, to-dos done). It only shows numbers your box really keeps.
• Cards. "Needs your yes" is now just Cards: the list of things waiting for your yes, behind the Cards count on the home. Yes still asks for your passkey, same as before.
• "Use my Claude subscription" is gone: Claude doesn't allow its subscriptions inside other apps. Your box switched back to its included brain; add an Anthropic key in Settings → Brain to use Claude.

Release 42 · 23 Oct 2026

• Your box hears about updates again. The website moved to usemagicbox.com, and on the way your box's check-in lost the key that says which box it is, so the website stopped telling it about new releases. Your box now follows the move and shows its key only to the website itself, never to any other address, and only over a secure connection. Nothing to do on your side.
• Grok and Muse as your brain. Settings → Brain → Use my own key now offers Grok (from xAI, key from console.x.ai) and Muse (from Meta, key from dev.meta.ai) next to Anthropic, OpenAI and OpenRouter. Paste the key, pick a model, tap Test it. It's your own bill, not your box's budget. If the key has no credit left, your box says so instead of calling the key wrong.
• Your own OpenAI key works when it was made with limited permissions. Your box used to call such a key wrong because it couldn't read OpenAI's model list; it now takes the key, and Test it says plainly if the key isn't allowed to chat. The model list no longer offers OpenAI models that only work in OpenAI's own apps (the "pro" and "codex" ones), and a busy OpenAI is reported as busy, not as out of credit.

Release 41 · 22 Oct 2026

• When "Use my Claude subscription" doesn't work, Settings → Brain → Test it now says why, in a short line for the maker: no saved sign-in, Claude refused the sign-in (and how long the saved sign-in is), Claude Code didn't finish in time, or the box's sandbox wouldn't start it. It never shows the sign-in itself. The same line goes in your diary. A refusal from Claude is now always reported as "not signed in", and a usage limit as a limit.

Release 40 · 21 Oct 2026

• Every page has its own chat. Pages → a page → ⋯ → Chat. Your box's page builder now talks about a page only there: its questions, the card, "v2 is ready", what failed. It no longer drops those into whatever chat you have open; the chat you asked from gets one line pointing you to the page's chat. Ask for a feature or a fix in a page's chat and it goes to that page; you don't have to name it.

• "Use my Claude subscription" now keeps working after you sign in. Claude's sign-in program sends its token in pieces, and your box sometimes saved only the first piece, so it looked signed in but every answer failed. It now waits for the whole token. If you signed in before, sign in once more.

Release 39 · 21 Oct 2026

• Connecting the Laso wallet with a key works. Your box now hands Laso the key the way Laso expects, and you can paste the key on its own, the "Authorization: Bearer …" line from Laso's dashboard, or Laso's whole "paste this to your agent" message: your box finds the key in it.

Release 38 · 20 Oct 2026

• Dark mode is readable. If your phone is set to dark, every screen of your box now has proper dark colours: text, cards, chat bubbles, buttons, the menu, the creature, the Memory cards and the locked screen are all easy to read. Nothing changes in light mode.

• "Use my Claude subscription" no longer gets stuck signing in after you paste the code. If Claude refuses the code, your box now says so straight away; if Claude doesn't answer, it gives up after a minute and a half and lets you start again.

• The chat box grows as you type, up to about half the screen, then scrolls. It shrinks back after you send. Shift+Enter starts a new line.

• The nightly read also makes to-dos and asks you about anything unclear. Things you agreed or were asked to do become to-dos (at most 5 a night), each saying where it came from; one you've dropped never comes back. When something matters and your box can't tell (is the dinner still on, which date is right), it asks you, at most 2 a night, and shows where the question came from. Your answer goes into your memory in your own words.

• Laso wallet (off until you switch it on). Settings → Laso wallet, with your passkey. Connect your Laso account (or make one on Laso's page), top it up by sending USDC on Solana or Base to the addresses it shows (with a QR), then just ask in chat: "Venmo Jo $40 for dinner", "get Mum a $25 Starbucks gift card", "make me a $5 card for my coffee". Every payment is a card that shows the fee and the total, and nothing is paid until you approve it with your passkey, whatever your other switches say. Your box keeps its own caps ($50 a payment and $150 a day to start; raising them needs your passkey) under Laso's. A card's number or a gift card's code shows only in Settings, on your passkey; your box's brain never sees them. There's no test mode at Laso: every payment is real money.

Release 37 · 20 Oct 2026

• Learning from your day is pickier, after its first real night. Someone who only booked one meeting with you no longer gets a page of their own. A weekly event is remembered once as a habit ("lifts on Saturdays at 12:30"), not once per week. It uses your own words for your people (your wife stays your wife), and it no longer writes facts about "an unknown contact".

Release 36 · 20 Oct 2026

• "Use my Claude subscription" no longer gets stuck on "Getting a sign-in link from Claude…". Claude moved its sign-in page to claude.com and your box was still looking for the old one.
• Pages that Claude or ChatGPT write go through the page factory too. Your box writes the checks the page must pass, asks you its questions, tests their page with your own notes and fixes what fails before it's under Pages. Nothing they send is installed as sent, however they send it. Nerds: BUGS 212.
• Your box learns from your day. Every night before it tidies its memory, it reads what came in and went out since it last looked: your email (every Google account), WhatsApp, Telegram, iMessage and your calendar. It writes down what's worth remembering (who people are to you, plans and promises, birthdays, what you like) and files it into your memory. It skips newsletters, adverts, sign-in codes, links and group-chat chatter, and it never keeps the messages themselves, only what it learned. Last night's card says what it read and learned, and every fact shows where it came from, with "I didn't mean that" to remove it. Switch each account on or off in Settings → Learn from my day, or forget everything it learned from one. It stops at $0.25 of thinking a night. Nerds: BUGS 213.
• When you connect something, it looks at the last month straight away, then tells you in chat what it learned and asks a few questions so it gets your people right ("Who is +1 512 555 0123? You message them most days").
• One small question a day, to get to know you ("Morning person or night owl?"). Never more than one, never a notification, always skippable; answer on Home or in chat. Switch it off in Settings → Learn from my day.
• Your people, matched to their numbers. A person in your memory can carry their WhatsApp, Telegram, phone and email, so messages read "Jo (your sister)" instead of a number. The box only links a number to someone when it's sure; otherwise it asks you. Numbers never go to the brain, and assistants you connect don't see them. Nerds: BUGS 214.
• The old "Morning catch-up at 7:00" line in what your box knows about you is gone, since the morning paper is off.

Release 35 · 20 Oct 2026 · UI overhaul

• UI overhaul. Your box has a new face and a new shape. No menu bar across the top: a round button at the bottom opens a menu that slides up, with Home, Chat, Memory, Cards, Connectors, Pages and Settings. Home is your box as a creature, with its mood on its face and one line about how it's doing; tap it when it needs you and the cards open. Cards are one line each with Yes and No, and unfold on a tap. Chat fills the screen and its messages scroll on their own. Memory opens on what happened last night, what your box knows about you, your maps, who uses your memory, and the things it set aside. Pages shows every page with its version and checks, what's being made right now, the questions your box has for you, and every older version to go back to. Settings and Connectors keep everything they had, in the new look. It's the same design you can try at https://mymagicbox.lovable.app/preview, now running on your own box with your own memory.
• Your memory no longer loses things. A line the nightly tidy couldn't place, or thought was small talk, used to vanish after three tries. Now it waits in Memory → Set aside with the reason, in full, and you can file it again or throw it away. A line you send back is kept for sure. The tidy also reads long lines whole, splits a line with several facts across the right notes, starts a new note when none fits, and keeps how you like to work. Nerds: BUGS 210.
• Chats come in sections: Home, Money, Family, Work, Health, or ones you make ("make a section for the house hunt"). After a couple of messages your box names the chat, puts it in a section and adds a few tags. Anything you set by hand stays as you set it.
• The morning paper and evening edition are off. Your box was in your face too much. The nightly memory tidy still runs at 3:00, and now you can see what it did: "Last night I filed 14 things into 6 notes and started 2 new ones (House hunt, Photography)." You can switch the paper back on in Rhythms.
• Cards fold to one line ("Send a WhatsApp to Sam", "Update to release 35"), with the full text one tap away. What you approve with your passkey is still the full text.
• Your box has a mood: happy, sleepy, thinking, needs you, hungry or not well, with one line saying why. Nerds: BUGS 211.
• Pages take their time now, like a real software team. Ask for a page and your box first writes what it's for and the checks it must pass, has those checks reviewed, and asks you a question or two if something is unclear. Then it designs the page, builds it, tests it against every check with your own notes, fixes what fails, and only then puts it under Pages, telling you how many checks pass. Each job is capped at $1.00 of brain spend and 20 minutes.
• Every change to a page is a new version that keeps what it was for, its checks and what changed. The old checks run again, so a change can't quietly break what already worked. You can go back to any of the last 10 versions, and compare two.
• New pages live under Pages instead of filling up the top menu. Pin one if you want it in the menu.
• Claude and ChatGPT can ask your box to make a page the same way (request_page), and see its questions; what they send is always a card.

Release 34 · 19 Oct 2026

• Telegram can be linked from Connectors → Telegram: type your number, then the code Telegram sends you (and your Telegram password, if you have one). Your box joins as one of your devices. It reads your chats, sends a message only after you say yes on a card, and only takes orders from your own Saved Messages, when you start with its name. Everyone else's messages are read, never obeyed. Reminders and routine results land in your Saved Messages too.
• Telegram needs the maker's Telegram app key. New boxes get it automatically; an older box picks it up on its next check-in with the website, and only while it has none. Nerds: BUGS 209.

Release 33 · 19 Oct 2026

• Send your box a link and it reads it: an article, a recipe, a listing, a menu. It no longer has to build a connector first just to open one page. It only opens links you send it yourself, never one it found in an email or on another page, and it tells you plainly when a site blocks it, needs a login or only works in a browser. Nerds: BUGS 208.

Release 32 · 19 Oct 2026

• The screens catch up with release 30. Settings now shows the five switches (What I may do), Lock my box and your recovery phrase, Updates (Ask me), Who can see my box, Settings → Brain with your own key or your Claude or ChatGPT plan, and every Google account you've added, each with Reconnect, Add Drive, Make main and Unplug.
• Cards have short titles (New page, Send as you, Save this?…), a "Sent by Claude" chip when another assistant sent one, and a Done label with Undo for things your box just did.
• Connecting Claude or ChatGPT: three checkboxes instead of four choices: Memory, My messages and mail, Can build for my box.
• The Skills tab is now Routines; starter pages from the library are under Pages → Add from the library.
• Update cards show what changes, with Update now, Not now and Skip this release.
• Removing the last passkey that opens your lock asks you first ("I have my recovery phrase"). Today shows how many cards are waiting, and when Today is hidden or replaced, a chip in the header still takes you to them.
• Flipping Safe mode, or any switch under Things I built, changes at once and no longer reloads the page or closes the section you're in. Nerds: BUGS 207.

Release 31 · 19 Oct 2026

• Claude and ChatGPT find their way round your box better: they're told what they may do, how your memory is laid out, how to build a page, and exactly what to tell you when something needs you (a box to tick, an account to connect, a locked box). Searching your memory for your own name now finds you.
• The new screens for release 30 were meant to ship here but weren't included; they come in release 32.

Release 30 · 19 Oct 2026

• Your box can lock itself so only you can open it. Its memory, accounts and files move onto a locked part of its disk, opened only by your passkey or a 12-word recovery phrase you see once. The people who make magic box can't open it, and neither can Google. A new box asks right after you set your passkey; an older box locks when you tap Lock my box in Settings → Safety.
• The moment you first lock your box, the makers' ways into the machine close: no remote login, no scripts from the cloud console, no serial console. Until you lock it they stay open, and its disk can be read by whoever runs the machine, so lock it.
• After the computer itself restarts, a locked box waits at a locked page until you open it with your passkey, and tells you when it restarted. Restarting your box from Settings, updates and crashes don't lock it, so the morning paper and the nightly tidy keep running. Claude or ChatGPT are told plainly that your box is locked.
• Updates wait for your yes. This is the last one that installs on its own: from now on each new version comes as a card, Update to release N, that shows what changes. Your box keeps doing its jobs whether you say yes, no or nothing.
• The website can't change your box's Google sign-in any more, and it now hands your box its keys once, when the box starts, sealed on the box instead of written into the machine's setup.
• Several Google accounts on one box: add your work account beside your personal one (Connectors → Google → Add another Google account). Your box reads mail, calendars and Drive from all of them, says which account each thing came from, and puts a draft in the right one. Unplug one, or pick the main one, next to it.
• Bring your own brain. In Settings → Brain, paste your own Venice, Anthropic, OpenAI or OpenRouter key and pick any model it offers, or sign in with your own Claude or ChatGPT subscription. It's your own bill, so the monthly budget doesn't cap it. If it fails, your box says so, and the included brain answers instead unless you switch that off. Your key is sealed on the box and never shown again.
• Five switches, and they all really work. Settings → What I may do: Change my box (Just do it, with Undo), Send as me, Share outside my box, Give other assistants access, Delete for good. No more modes. Ask for a page, a look or a routine and it just happens; sending, sharing, giving access and deleting still ask first. After your box has read an email, a file or a web page, anything that sends, shares or deletes is always a card.
• One word for your own pages: Pages. The Skills tab is called Routines, and ready-made pages from the library are starter pages, under Pages. Cards have nine plain titles, and a card another assistant sent says who sent it.
• Keys for other assistants have three checkboxes: Memory, My messages and mail, Can build for my box. Tick any of them; reading your memory is always in.
• The manual your other assistants read (/for-bots) now starts with how to connect, every tool in one line, and how to build you a page end to end, and says how to use your box as one shared memory.
• Some of the screens for this (Lock my box, Brain, the list of Google accounts, the switches and the three checkboxes, the Routines name) arrive with the next release of the box's app; the box's side is all in this release.
• Nerds: BUGS rows 125 to 140; docs/SITE-LOCK.md, docs/UI-LOCK.md, docs/UI-BRAIN.md, docs/UI-MULTI-GOOGLE.md, docs/UI-SIMPLE.md.

Release 29 · 19 Oct 2026

• Your screen is yours to arrange. Ask your box for a hamburger menu, fewer tabs, a different order or different names, what opens first, or one of your pages as a tab or as the home screen. It becomes a card (or just happens, if you set "Arrange my screen" to Just do it in Settings → What I may do), and undo puts it back. Settings always stays. Claude or ChatGPT can send you an arrangement too. The screen itself follows in the next release of the box's app.
• A page can be your menu, your hub or your whole front screen: it can now open your other pages and tabs, read and tick your to-do list, see which cards are waiting for you, and read the diary.
• Your box can build itself a connector that reaches any website, when you say so on its card. That is how it reads a link, a video, a repo or a tweet you send it: it builds the reader, you approve it once.
• The box's standing order has changed: when you ask for something it can't do, it builds it (a page, a connector, a layout change, a look change) instead of saying no.
• Nerds: BUGS rows 122 to 124.

Release 28 · 19 Oct 2026

• Pages no longer add tabs to the top menu. The box keeps its seven tabs; every page is listed under Pages and opens in place with a back button. A page that is off still shows there with its status, and opening it takes you to Settings → Things I built. From Rava's owner report.
• The icon you asked for in chat now shows by the box's name and as the browser tab icon, on real boxes and in the demo.

Release 27 · 19 Oct 2026

• Pages work on iPhones. Safari treated a page's calls to the box as forbidden, so no page ever showed data or saved anything there; the box now names its own address in the rule that allows those calls. A page you already built needs one more open.
• The chat window has a fixed height with its own scroll bar, so the box to type in stays at the bottom and the page no longer grows forever. New messages scroll into view unless you had scrolled up. The first fix from an owner's "Tell the maker" report.
• Ask for a different icon ("make your icon a little box") and the box changes the mark by its name and the browser tab icon.
• Pages live under the Pages tab; the box says "under Pages" instead of "in your menu".
• WhatsApp linking by code is confirmed working on a phone.

Release 26 · 19 Oct 2026

• Settings → Other assistants offers the new kind of key, "Can read my messages and mail", and shows it in the connections list.
• Things I built and the Pages tab say when a page reads all your memory and which accounts it uses.

Release 25 · 19 Oct 2026

• "Remember that…" just remembers. When you ask in your own words to remember, note, save, add to your list or remind you, the box writes it, even in a turn that read email or messages. A card still comes only when a fact surfaced out of outside content without you asking.
• Your other assistants can read your messages and mail. A new choice when connecting Claude or ChatGPT, or a new kind of key in Settings: "Read and write, and read my messages and mail". That connection can read your WhatsApp, iMessages, email and calendar through the box, never send, and every read is in your diary. Claude, ChatGPT or an assistant on your own computer only.
• A page can see all of your memory when its card says so, and it can change or forget facts in the notes it reads. Your Undo brings any line back.
• A page can use your accounts when its card lists them: it reads your WhatsApp, iMessages, mail, calendar or Drive through the box. Sending from a page is always a card.
• Nerds: BUGS rows 116 to 118.

Release 24 · 19 Oct 2026

• This page. Every release from now on says what changed, here and at /changelog on the site, and Settings → About on your box links to it.
• A release cannot be signed without its entry here.

Release 23 · 19 Oct 2026

• Apps and sections are one thing now: a page. Every page lives in your menu, keeps its own notes in your memory and reads the maps you allow, so a page and your chat know the same things. Pages you built before were moved over with their data. The Apps tab is now Pages; Things I built lists Connectors, Pages and Looks.
• A new page is in your menu the moment it is built. The Activate step is gone. If a page breaks the first time it opens, the box takes it out and says so.
• If a page opens but never talks to the box, the diary now says so and names the cause.
• WhatsApp linking: one code per attempt, and asking again shows the same code for about ten minutes. The sheet counts down. A half-finished attempt can no longer spoil the next one, and the diary says when your phone accepted the code, when it ran out, and why an attempt stopped.
• Every page remembers who made each version.
• Pages have a documented address (/s/your-page/ on your box, your own sign-in needed).
• A connector that never leaves the box no longer has to name a website.
• When the nightly tidy drops a line it couldn't file, the diary names it and why.
• The sandbox's own security warning is gone from connector test logs.
• From Lili's Claude's report: docs now say the sign-in limit is for connectors builders write, not the box's own Google connector.
• Nerds: BUGS rows 105 to 115.

Release 22 · 18 Oct 2026

• Chat threads for everyone: a thread bar with General and + new, and each thread can be renamed, cleared or deleted. Memory is never touched by any of those.
• Tell the maker: a feature request, a bug or a question, from chat or from Settings. You see the exact report first; it carries your words and the box's own status lines, never your memory, chat, messages or mail. The makers reply by email.
• The WhatsApp sheet leads with the laptop route: open your box on a laptop or tablet, show a QR, scan it with your phone.
• Nerds: BUGS rows 101 to 104.

Release 21 · 18 Oct 2026

• Google connect no longer ends on "bad_origin".
• The iMessage helper installs cleanly (the Mac's security tool was being sent a command it doesn't have). A fresh install line and code are needed.
• WhatsApp: the box now announces itself as a Mac companion, which WhatsApp accepts; a failed link writes WhatsApp's own reason in the diary.
• Nerds: BUGS rows 98 to 100.

Release 20 · 18 Oct 2026

• No more invite codes. Asking for a box carries a private note (who you are, what you want help with) that only the makers read.
• Private-beta copy wherever someone asks for a box; setup says "picked", never "connected", until a box exists.
• Voice notes are described as what they are.
• Nerds: BUGS row 97.

Release 19 · 18 Oct 2026

• Updates that need something installed on the machine now do it themselves, so no feature ever needs a Rebuild (Rebuild erases a box; Restart keeps everything).
• Nerds: BUGS row 96.

Release 18 · 18 Oct 2026

• The box builds itself: connectors in a sandbox that reach only the websites on their card, sections (now pages) with their own notes, and changes to how it looks. Every one starts as a card you approve with your passkey.
• Your other assistants can build for it: a key that "Can propose changes to your box" lets Claude, ChatGPT or Codex send a connector, a page or a look change; each arrives as a card that says who sent it.
• Google Drive, read-only, as an add-on to Google. iMessage from your Mac with a small helper.
• Three security reviews fixed; Google sign-in is bound to the device that started it.