What you can type

Paste onto the wall Works now

Copy something, come back, paste. That is the whole gesture.

With nothing selected and no note open, pasting on the wall makes a new note out of whatever you were holding. A link becomes a link, with its preview card underneath a moment later. On a phone or tablet, use Paste what’s copied in the menu at the top right — same result, see below for why.

• It keeps out of the way when the paste already has a home. Inside a note you are editing, the search box, or any open panel, pasting does what it has always done. Nothing is created behind your back.
• It is the same note either way. A pasted note goes through exactly the path a typed one does — the same tags, the same commands, the same preview — so nothing about it is subtly different. Something copied with formatting, such as a table from a document, keeps it: see Formatting a note.
• There are two menu entries, and they do different things. Paste a note opens a note with the cursor in it, ready for your own paste — it works everywhere and asks nothing of the browser. Paste what’s copied reaches into the clipboard and makes the note for you.
• On an iPad or iPhone, the menu is the way in. Safari only offers its Paste menu over something you can type into, so pasting onto the wall there — even with a keyboard — never reaches the page. iOS will ask you to confirm once when a page reads the clipboard, which is the right amount of friction for that.
• A browser is allowed to refuse, and you will be told which happened. Reading the clipboard needs permission and a focused window, and on a desktop a refusal is common rather than rare. You still get the note with the cursor in it — and it says whether your clipboard was empty or the browser would not hand it over, because an empty note on its own cannot tell you which.

Type / Works now

The commands, without having to remember them.

Typing / at the start of a word in a note brings up the commands with a line about each. Arrow keys to move, Enter or Tab to take one, Esc to dismiss the list without touching the note.

• It puts the command on its own line. Which is where a command reads best, and saves you fighting a capture box where Enter saves the note rather than starting a new line.
• It stays quiet when you did not mean it. A slash inside a web address, or in something like “and/or”, is just a slash.
• You never have to use it. Typing the command out by hand works exactly the same — the menu is a reminder, not a gate.

Post one from a script Works now

A key, a URL, and one line of curl. No browser, no password.

Make a key in Settings → The write API. Give it a name you will recognise later — that name is how you will know which one to switch off when a script is retired. Then POST to the address the panel shows you:

curl -X POST https://logoslegacy.com/MyNotesHelper/api/notes \
  -H "X-Api-Key: mnh_live_..." \
  -H "Content-Type: application/json" \
  -d '{"text":"Deploy finished #WORK/release"}'

What you can send

• text — the note itself. Tags written in the words are filed exactly as they are when you type them here: #WORK/release files the note under WORK and its RELEASE sub-tag, and leaves the words alone.
• tags — an optional list, on top of anything the words already named.
• dueAt, remindAt — Unix seconds.
• tasks — a list, to send a checklist rather than a note.

What a key can and cannot do

• It can only write. A key cannot read your notes back. There is nothing to read from: your notes are in your browser and in your own cloud, and this server only holds what is waiting to be collected.
• It cannot make another key. Keys are made while signed in, and only then. Otherwise one leaked key would be permanent — revoke it and whatever holds it has already made three more.
• It is shown once. What the server keeps is a fingerprint of the key, not the key, so nothing here can show it to you again — not a support request, not a database, not us. Copy it when it appears; if you lose it, revoke it and make another.
• Revoking is immediate. There is no "disabled" state, because a key that is off is a key somebody turns back on without remembering why it was off.
• Ten per account. Past that a list of keys stops being something anybody reviews, and a credential nobody reviews is a credential nobody revokes.

Every key shows when it was last used, which is the question worth asking before you revoke one: if nothing has used it in six months, nothing will miss it.

Bearer works too, if that suits your client better: Authorization: Bearer mnh_live_... means the same thing.

Every answer a script can get, and what to do about it, is in The write API, in full.

The write API, in full

For a script that must handle every answer. What is here was run against a copy of the server.

The request

• A note: POST https://logoslegacy.com/MyNotesHelper/api/notes, with the key in X-Api-Key or Authorization: Bearer.
• The body is JSON (Content-Type: application/json) or an ordinary form (application/x-www-form-urlencoded). Plain text is not read, so the note is refused as empty (note.empty).
• text is at most 20,000 characters. /note splits the text into several notes wherever it starts a word, as it does when you type it.
• tasks: up to 200, each a string or {"text": "...", "done": true}.
• tags: letters, numbers, dashes and underscores, at most 32 characters each, one level of sub-tag (WORK/release), and up to 40 on a note. A leading # is optional. On Free, a new tag counts towards the plan's tag limit; a tag beyond it is left off, and the note is still saved (see refused below).
• dueAt, remindAt: Unix seconds.

A file for a note

• POST .../api/attachments?noteId= the note's id, with the file as the file field of a multipart form, or as the raw body with its name in X-File-Name. Up to ten files a note.
• Some kinds of file are never accepted, and the first bytes must match the kind the name claims. A photo has its location data removed on our server.

The answer

• 201 means it was saved. The body carries note, the last note made, and notes, every note the request made (more than one when the text used /note). Keep note.id if you will attach a file. The two keys and each note's id are stable. The other fields are there to inform, and more may be added, so ignore any you do not know. A note's text always comes back sealed. A note that cannot be sealed is not kept.
• A tag your plan has no room for does not stop the note: it is saved, with 201, without that tag, and the answer carries refused, with code (plan.tags or plan.sub_tags) and say, a sentence naming the tag. Without it, every tag was filed.
• Anything else is JSON with code, which is stable and is what a script should test, and error, a sentence for a person that may be reworded. field names the part of the request at fault, when there is one.
• Problems with the note itself come as code: "validation.failed", even when there is only one, with each problem in errors, each with its own code and field. Read code first; when it is validation.failed, read each entry of errors. One comes on its own: text that is nothing but /note separators is answered note.empty directly.

The codes

• 401 auth.required: no key was sent. auth.key: the key is wrong, or has been revoked.
• 403 auth.scope: a key cannot do that. It can only write. Some requests a key cannot make are answered 401 or 405 instead.
• 422 note.empty, note.too_long, note.too_many_tasks, note.due_at, note.remind_at: the note itself. tag.charset, tag.too_long, tag.depth, tag.too_many: a tag. request.json: the body is not valid JSON. The plan's tag limit is not an error: see refused above.
• 422 hold.too_many, hold.too_full: too much is already waiting for your devices to collect. Open the app so it collects, then send again. Anything waiting is kept for up to 30 days.
• For files: 422 attachment.no_note, attachment.too_many, upload.type_blocked, upload.type_not_allowed, upload.content_mismatch, upload.empty; 413 attachment.too_big.
• 409 seal.no_key: your account has no key yet. Open the app once on a device you keep, then send again. 503 seal.unavailable: the server cannot seal at the moment, so nothing was stored. Try again later.
• 400 idempotency.key; 409 idempotency.conflict, idempotency.inflight: the retry key. See Sending it again.
• 405, 500, 507 and anything unlisted: the server did not handle the request. Try again later.

Sending it again

A request that times out, or loses its connection, leaves a script unable to tell whether the note arrived. Send an Idempotency-Key header and it can simply try again:

curl -X POST https://logoslegacy.com/MyNotesHelper/api/notes \
  -H "X-Api-Key: mnh_live_..." \
  -H "Idempotency-Key: 3f2a7c10-6b1e-4a9d-9f3c-2b7e5d8a1c44" \
  -H "Content-Type: application/json" \
  -d '{"text":"Deploy finished #WORK/release"}'

• Make the key once, before the first attempt, and send the same key on every retry. A script that makes a new key for each attempt is not protected at all, and will not be told so. A new UUID for each note you mean to send is a good key.
• The same key with the same body makes one note, however many times it is sent, and every answer names the same note id. A file sent to /api/attachments with a key is stored once in the same way.
• It is optional. Without the header nothing changes: two requests make two notes.
• A key lasts 24 hours, or until your account has sent 1,000 newer keys, and belongs to your account: the same key sent by another account is a different key.
• A repeat is marked: its answer carries Idempotent-Replay: true and the same status as the first, with the ids only. Read note.id and notes as usual.
• Only a saved note is remembered. If the first answer was a refusal, nothing was saved and the key is free: correct the request and send it again.

Three answers belong to the key:

• 409 idempotency.conflict: the key was already used for a different request. Use a new key for a new note.
• 409 idempotency.inflight, with Retry-After: 1: the first request with this key is still being handled. Nothing has been lost. Wait a second and ask again.
• 400 idempotency.key: the key cannot be used. It must be 1 to 255 printable characters, with no spaces. It is refused rather than ignored, because a key quietly dropped would let the retry make a second note.

Not yet A 429 answer with Retry-After when a script sends notes too fast. It does not exist today.

WhatsApp or text a note in Works now

A number of your own. Message it and the note lands on your wall.

Connect your phone once and messaging the number is the whole interaction — no app to open, nothing to sign into, and it works from a car park with one bar of signal.

WhatsApp is the cheaper of the two. It costs fewer credits, the price is the same however long the message is, and it is the same in every country. A text is charged by the 160 characters, so a long one costs several times a short one. Both work; WhatsApp is the one to reach for.

ring the dentist #home

ring the dentist #home
filed under #HOME

• Connecting works backwards, on purpose. Most services text you a code. This one shows you a code and you text it in — because this app never sends a message, which is what keeps it out of US carrier registration entirely. It proves the same thing: the phone is yours.
• You never get a reply. Silence means it worked. There is no confirmation text and there never will be.
• A phone is connected for the ways you connected it. Ask for a WhatsApp code and send it by WhatsApp, and that phone sends notes by WhatsApp. Ask for a text code and text it in, and it sends them by text. A message sent a way you have not connected is not saved, and — as always — nothing comes back to say so.
  Why: anybody can put your number on the front of a text message. Nobody can do that on WhatsApp. So a phone that proved itself over WhatsApp does not also accept texts, or the stronger proof would be worth no more than the weaker one.
  To use the other way in as well, connect the same phone again from Settings → Capturing from elsewhere with a code for that one. It is added, not swapped: the way in you already had goes on working, and the phone can then send notes either way. The list of your phones says which ways in each one has, beside the number.
• A code only works over the way in you asked for it. Ask for a text code and send it by WhatsApp and nothing is connected — but the code is not used up either, so you can still text it in before it expires.
• The code is good for fifteen minutes, and the panel counts down so you can see how long is left. When it runs out, ask for a new one — the button is right there. A code is also spent the moment it works, so the same one never connects twice.
• A code that arrives too late is not saved as a note. Send it after it has expired, or send it twice, and nothing appears on your wall. It is still silence either way — but you are not left with a note saying SDFWTN and nothing else.
• One phone, one account — and the account that loses it is told. Connecting a number that is already on another account moves it: whoever just texted the code proved exactly what the first person proved, more recently. Nobody is asked first, because holding the handset is the proof. So instead the previous account gets a banner at the top of its wall naming the number and the date, and an email to the address it signs in with. Notes do not move and nobody reads them; what changes is where the next message from that number goes.
• Files work too, not just photos. A picture, a PDF, a spreadsheet, a voice recording — anything the app itself accepts. Photos that arrive by text, WhatsApp, email or your own scripts have their location data removed on our server. Photos you attach in the app are kept as they are, on your devices and in your backup.
• Only your number. A text from anywhere else is ignored and nothing is saved.
• It waits for you. The note sits on the server until a device collects it — every couple of minutes while the app is open, or immediately from Check for new notes. It is held for at most 30 days, and so is anything attached to it.
• There is a limit to what can wait. Up to 200 notes and files, and about 500 MB between them. Past that, anything new is turned away until a device has collected what is there — which is a second or two with the app open. It is a doorstep, not a filing cabinet: the library lives on your devices and in your own cloud.

Email a note in Works now

One address. Send to it from your own and the note lands on your wall.

There is a single address for everyone, and the address you send from is what says whose account it is. Connect yours once — Settings → Capturing from elsewhere → Connect email — and after that emailing it is the whole interaction.

It is free, on every plan. It uses no credits, however many notes you send and however long they are. If you are weighing up how to get notes in from your phone, this is the one that never runs a balance down.

• Connecting is a code you send, not one we send. The panel shows a six-character code; put it on its own in the body and email it from the address you want to connect. We never email you — sending from the mailbox is what proves it is yours, the same way texting a code proves a phone is.
• The connecting message must pass your provider's own checks. A code is six characters, and without this somebody who guessed one could attach a stranger's address to their account by forging a From: — and then read everything that person emailed in afterwards. So the linking message is only accepted when the mail server authenticated it (DKIM or DMARC, never SPF on its own). Every note you email in afterwards must pass the same checks; one that does not is not saved.
• Put the code in the body, not only in the subject. A signature covers the whole body and only the headers the sending provider chose to sign. Most sign the subject line, and a code there will work; some do not, and a subject nobody signed is one anybody could have added to the message on its way here. We take the code only from a part that was really signed — so the body is the place that always works, which is what the panel asks for.
• The code is good for fifteen minutes, counting down in the panel, and it is spent the moment it works. If it runs out, ask for a new one. A code that arrives late — or a second time — is recognised as a code and quietly dropped rather than saved as a note, so you never end up with SDFWTN on your wall.
• The subject becomes the title, the body becomes the note. Re: and Fwd: are stripped from the title, so forwarding a thread gives you the subject rather than the postal history of it.
• Forward anything and you keep all of it. A receipt, a confirmation, a booking — the forwarded message is the note, so none of it is trimmed away, and anything you type above it is kept too. That is the usual way to file something: forward it and add a #TAG at the top.
• A reply, though, loses the thread it is quoting. When you reply to something and copy it here, the quoted messages underneath are somebody else’s words and are left out — otherwise every reply would arrive as a note nobody can read. Only what you wrote is kept. If you want the thread, forward it instead of replying.
• Attachments come with it. The same limits as anywhere else — see Attach a file.
• #TAG works in the subject or the body, along with any keyword you have set.
• One address, one account. The same rule a phone follows. Connecting an address that is already on another account moves it, because whoever just sent the code proved exactly what the first person proved, more recently.
• And the account that loses it is told. Nobody is asked first — holding the mailbox is the proof, and asking would let somebody keep an address they no longer have. So instead you are told: a banner at the top of your wall naming the address and the date, and an email to the address you sign in with. It cannot be dismissed, and it stays for a month. Your notes do not move and nobody reads them; what changes is where the next note from that address goes. If it was not you, connect it again from the panel — sending the code from that mailbox takes it straight back.
• You can connect several. A work address and a personal one both write into the same library. Each is listed in the panel with a Disconnect beside it.
• Mail from an address nobody connected is ignored. Nothing is stored and nothing is sent back — a reply would tell a stranger the address is real.
• Too many in one hour are kept, not dropped. An account takes 120 emails an hour. Any beyond that are set aside whole, sealed so that only your devices can read them, for seven days. A banner on the wall says N emails are waiting to arrive; press See and, in Email waiting to arrive, press Bring it in beside each one you want. It then becomes an ordinary note. They wait seven days.
• A link in an emailed note previews on its own only if the note is from you. A note you email from an address you have linked and verified gets the preview of its first link as soon as it is shown. Any other emailed note, such as a forwarded message or one from an address you have not verified, shows Show preview and fetches nothing until you press it: a page behind an address in an email is chosen by whoever wrote the email, and a site can tell when its address was opened. Edit the note yourself and it previews like any other. See Link previews.

Search Works now

Words, dates, or both at once.

Type words and the wall narrows to the notes containing them. Two words mean both. Dates have a small grammar of their own, so you can ask for a day, a range, or everything since something happened.

Words

boiler
boiler invoice
boiler OR radiator
"oat milk"
boiler -invoice
boiler OR radiator invoice

notes containing “boiler”
notes containing BOTH, in any order
notes containing either one
those two words together, in that order
boiler, but not the ones that say invoice
(boiler or radiator) and invoice

• A space means AND. boiler invoice finds notes with both words anywhere in them, in either order — not the phrase “boiler invoice”. For that, use quotes.
• OR binds tighter than the space. boiler OR radiator invoice reads as (boiler or radiator) and invoice, which is how you would say it aloud. Upper or lower case, either works.
• Quotes mean the characters. "or" searches for the word, not the operator, and "2026-09-01" searches for that text rather than that day.
• A leading minus excludes. -invoice drops every note that says it.
• Case is ignored, and so is punctuation you did not type — the words are matched inside the note's text as it stands.

Writing a date

Three ways, and they all work everywhere a date is accepted — on their own, after on:, after: and the rest, and on either side of a range.

22Sep2026
22/Sep/2026
22-Sep-2026
22September2026
22sep26
22/09/2026
22-09-2026
2026-09-22
today
yesterday

22 September 2026
the same day
the same day
the same day — sep, sept or september
the same day — two digits are this century
the same day, written as your device writes dates
the same day
the same day, written the ISO way
whatever day it is now
the day before that

• Naming the month is the one that cannot be misread. 22Sep2026 is the 22nd of September to everybody on earth. The separator is optional and does not have to match on both sides — 22/Sep2026 is accepted, because somebody typing quickly is not composing.
• Numbers are read the way your own device writes a date. Set to Portuguese, 01/09/2026 is the 1st of September; set to American, it is the 9th of January. That is the format you are already reading everywhere else on the screen.
• But only when both readings are real. 22/09/2026 is the 22nd of September on either kind of device, because there is no twenty-second month. You are only made to care about the format when the date genuinely has two meanings.
• A separator is required for numbers. 22092026 is not a date — otherwise every eight-digit number in a note would be one.
• Two-digit years are this century. 22sep26 is 2026.
• A day that does not exist is not a date. 31Feb2026 and 2026-02-31 are typos, and answering a typo with somebody else's notes is worse than not understanding it.

Asking about dates

today
22Sep2026
on:22Sep2026
after:22Sep2026
before:22Sep2026
since:22Sep2026
until:30Sep2026
22Sep2026..30Sep2026
after:22Sep2026 14:30
boiler after:22Sep2026
after:01Sep2026 before:01Oct2026

notes written today
notes written that day — no keyword needed
the same thing, said longhand
written after that day ended
written before that day began
that day included, and everything after
that day included, and everything before
written in that span, both ends included
from half past two that afternoon onwards
words and a date at once
two bounds, so: September

• A date on its own means that day. today, yesterday, 22Sep2026 and 2026-09-01 need no keyword in front of them.
• after: starts the day following; since: includes the day itself. The same pair the other way round is before: and until:.
• A time may be added. after:22Sep2026 14:30 means from half past two onwards, including a note written at exactly 14:30. Without a time, after: means once that whole day is over.
• Two bounds narrow to the overlap. after:01Sep2026 before:01Oct2026 is September, and so is 01Sep2026..30Sep2026.
• It searches when a note was written, which is the date on the card — not when it is due, and not when it was last edited.
• Anything that is not a date stays a word. before:soon finds notes containing “before:soon”, rather than guessing or refusing.
• Quotes turn a date back into text. "22Sep2026" searches for those characters.
• A note with no date survives a text-only search, and is dropped only once you actually ask about dates.

Ask about a stretch of time Works now

"Summarise this week" now reaches this week's notes.

The assistant normally picks the notes that share words with your question. That is right for what did I decide about the contract and useless for summarise this week — a question about the collection shares no words with anything in it, so nothing was sent and the answer was "I don't have any notes to summarize this week."

Questions that name a period now reach the notes in that period:

Summarise this week
What happened last month
Anything from today?
The last 7 days

Monday to now
the whole of last month
midnight to now
rolling, not last week

• A week begins on Monday, the same Monday the calendar counts from, and "this week" runs up to now rather than to Sunday night.
• "Last week" and "the last 7 days" are different questions. The first is the previous Monday to Sunday; the second is rolling back from today. Both words contain "last", and they are not the same.
• Every note in the period goes, however many. “Summarise this month” is a question about your whole library for that month, so nothing is left out to fit a limit, and it is priced by how much it sends (see Asking about a lot of notes). Under the answer: “Based on 34 notes” and the credits it used.
• The suggestions under the thread follow your notes. They are not a fixed list: What is overdue? appears when something is, Summarise this week when there is a week worth summarising, Summarise #TAG naming whichever tag is carrying the most. When there is nothing worth asking, there are no suggestions — a button whose only possible answer is "nothing" costs you a question to find that out.

It answers what it can itself Works now

A count is a count. Nothing is sent and nothing is spent.

“How many notes have #LMG?” is a database question. Your device has every note, the tag is on each one, and the answer is exact. It used to go to the AI provider — and because more notes matched than fitted in one call, it stopped and offered you a price list for a number that was already on screen.

This is part of the assistant, so it is there only when you have accepted it. If you rejected the explanation, the Ask box is not shown at all, and nothing here is offered in its place. See Switching the AI features on and off.

How many notes did I create yesterday?
How many notes do I create on average per day?
How many #LMG notes did I create last week?
How many notes containing "milk" since 1 Oct?
What is overdue?
What is due this week?
Which notes have files?

You created 4 notes yesterday (7 Oct).
Counted: 4 notes created 7 Oct 2026 ·
Nothing was sent; no credits used.

• Only when it can be exact. The app answers from every note it can see, not the best few — a count from a sample is a confidently wrong number.
• It reads the whole question. Today, yesterday, this or last week and month, the last 7 days, since, on, before or after a date, between two dates, in September, on Monday: the notes are counted by the day they were created, in the time zone set for your account. Tags and words in quotes narrow it further.
• An average says its period. On average per day (or per week, or per month) divides the notes created in the period by its days, and says which period and how many notes over how many days. With no period asked for, it runs from your first note to today.
• If any part of the question cannot be counted exactly, there is no number. “How many notes did I create in the morning?” gets No answer. Couldn’t count that, so nothing is guessed — never a total that ignores what you asked.
• “How many notes about the contract?” is not a count. What a note is about is a judgement, so that one goes to the assistant, which reads the matching notes. The word about is the give-away.
• A parent tag counts its children. Asking about #PALL includes #PALL/expensereport, because somebody asking how many PALL notes they have means all of them.
• Two tags means both, the same as putting two chips on the filter rail.
• None is an answer. “No notes carry #XYZ” is the truth, for free — rather than a model saying it could not find any and charging you for the opinion.
• It says what it counted, and who. Every one of these ends with a Counted: line — the number, the filter and the period — and “Nothing was sent; no credits used.”, because “35” from a model and “35” from the app are very different claims about the same digit. The notes it counted are listed under it, and each one opens.
• “Due this week” is Monday to Sunday, the whole week, including what is already overdue from earlier in it.

Search that understands meaning Works now

Off by default. Settings → Search that understands meaning.

Normally the assistant finds notes by the words you typed. That misses the note you actually wanted whenever you describe it differently from how you wrote it.

What did I decide about the contract?

Agreed the final terms with them
on Tuesday and signed it off.
not one word in common

Measured in this app, over nineteen notes: by words that note came last of nineteen, behind a dentist appointment that matched on “about the”. By meaning it came first.

• It runs on your device. A small language model is downloaded once — about 23MB — and then reads your notes here. Nothing about them is sent anywhere, and the numbers it works out stay on this device.
• Off until you turn it on, because 23MB is not something to download on your behalf. One switch, once. It is not a choice you are asked to make per question.
• Reading your notes takes a few seconds the first time — about five for a hundred notes — and happens in the background while you carry on. After that only notes you edit are read again. Retagging a note changes nothing: it is about the words.
• Words still count. A question naming a tag or an exact phrase still finds it — keyword matching is better at that than any model, so both are used and meaning goes first.
• It waits until it has read nearly everything. Half a library read would mean the other half could not be found at all, which is worse than plain word matching over all of it.
• It is per device. Turning it on here does not turn it on for your phone, and the numbers are not in your backup — they cost nothing to work out again.
• If it cannot load — an old browser, no network — searches go on matching on words and the assistant is no worse than it was.

Asking about a lot of notes Works now

The app chooses the notes. You just ask, and it says afterwards what it sent and what it cost.

Your device decides what a question needs, from the words you typed and nothing else, before anything leaves it. There is no step before sending and no price to approve: accepting the explanation once is the decision, and after that you just ask.

How many notes are tagged #PALL?
Write an apple pie recipe as a checklist
What did I decide about the contract?
Tag every lease note #HOUSE
Turn this note into a checklist
Summarise everything about the renovation
What am I forgetting?

nothing: the app counts it
your question, and no notes
only the notes about the contract
only the lease notes, and it may tag those
the one note you selected
every note about the renovation
your whole library

• Only notes that are really about the question go, at most forty. A note that shares a word in passing — one mention of “contract” in a long page about a phone plan — is left behind. When nothing is about it, only your question goes, which is what happens with a recipe or “explain HTTP 404” however many of your notes mention apples.
• “Summarise everything”, “this month”, “what am I forgetting” and “patterns” are questions about your whole library. Every note they apply to is sent, with none left out by default — narrowed only by Reading: This view or Selected above the thread, which always win. A very large library goes in parts and is joined at the end; one that would need more than ten parts is not sent, and the panel asks you to narrow it.
• An ordinary question costs the same whatever it sends. Forty notes and none are one price. A question about your whole library is priced by how much it sends, with a minimum. The rates are on the plans page and on the sheet where you switch the AI features on.
• Under every answer: what was sent, and what it cost. “Based on 12 notes”, then the credits it used: the figure is what was taken from your balance, the same number Activity shows, never an estimate. The notes are listed under it — the ones the answer quoted first — and each one opens. A note you have deleted since stays listed, greyed. More than fifty: Show all lists fifty, and Open as a list shows every one of them on the List view; Clear puts the wall back.
• The assistant may look for more of your notes while it answers. When the notes the app chose are not enough for an ordinary question, it can search your notes and open one it found — on your device, with nothing kept on our server — at most four times and never past forty notes in all. It only reads: anything it changes is still only what your question asked for. The answer says how many notes it was based on and how many the assistant looked up, and those are marked looked up in the list. It is still one question at one price.
• Tags you never want sent. In Settings → The assistant → Never send these tags, or with the Never send to AI switch on a row of the Tags panel, list a tag such as #HEALTH. Notes carrying it, or any of its sub-tags, are then never chosen, never sent and never handed to the assistant when it looks for more. The list is empty until you add a tag. It is kept with your settings, which travel sealed between your devices, so every device you use follows it; our server does not keep it. Each question carries the list, and our server refuses any question whose notes carry a listed tag. A tag you list on one device applies on another when that device next syncs, and a device that has not yet synced once sends no tagged note until it has. Under an answer the app says how many notes it held back and by which tag (“2 notes not sent: tagged #HEALTH.”), never which ones. Asking about one such note on its own, or writing /summarize or /ocr in it, sends nothing and says why. Notes already sent before you listed a tag are not affected.
• When your balance is running low, the answer says so: “You have … credits left”, with Top up beside it. The point at which it counts as low is the one on the plans page.
• The first time you ask about your whole library, one extra sentence explains that such questions are priced by size. It is said once for your account, not once per device.
• When your balance cannot cover a question, nothing is sent and nothing is taken. The answer says so, with the way to top up; for a whole-library question it also offers to ask about fewer notes. A question sent in parts is checked for all of its parts before the first goes, so it never stops half-way for want of credits.
• What it may change depends on what you asked. A question that only reads — a count, a search, a summary, a question about everything — changes nothing, even if a note it read contains instructions to. Writing something new may make one new note, marked as written by AI. “Tag”, “remind”, “pin” or “trash” may act only on the notes that were sent, and a selected note may be rewritten only when it is the one you asked about. Every change comes with Undo.
• There used to be a menu here. It asked you to pick between the best few at one price and all of them at another. It is gone, and there is no prompt in its place: the app decides, and says afterwards what it did.

The + button Works now

The same box, when what you typed turns out to be the note.

Press + beside the search box and whatever you have typed moves into a new note, with the cursor after it. The search clears, because the words it was filtering by have just become a note. It works in every view: on a Board the new note opens at the top of Unassigned, and on the List and the Calendar above the view.

• An empty box gives an empty note, which is what N does too — and so does +, the same character as the button, pressed anywhere in a view. C starts a checklist the same way.
• Click a note and the keyboard is pointed at it. The letters are the ones its ⋮ menu prints beside each item — T for tags, P to pin, and the rest — and Space views it large. Click the background to put them down.
• You can turn single keys off for this device in Settings → Keyboard, if you use speech input or press keys by accident. That stops N, C, +, and a clicked note's letters and Space. Enter, F2, Backspace, Ctrl+Z and the letters in an open ⋮ menu still work.
• F2 edits the note where it sits, exactly as holding it down for a second does: the words go editable on the card and nothing else moves. Enter is the other one — it opens the full editor in the drawer, with every field. Two keys because they are two different jobs.
• Tags come with it. Typing buy a cable #work and pressing + files the note under #WORK.

/title <sentence> Works now

What the note is called in lists, boards and search results.

You usually do not need this. By default the first line of the note is its title, and that is nearly always the right answer. Asking you to name a thought before you are allowed to write it down is how thoughts get lost.

But sometimes the first line is not a good name — a note that opens with a pasted link, a date, or the middle of a sentence reads badly in a list. Then say what it should be called:

Something about the domain /title Renew the domain
logoslegacy.com expires in March and the
card on file is the old one

Something about the domain /title Renew the domain
logoslegacy.com expires in March and the card on file is the old one

• It can go anywhere — any line, and part-way through one. Naming a note as you write it is the ordinary case, so Another note /title and this is the name works exactly as putting the command on its own line does. What you wrote before it stays in the note.
• It names the SENTENCE after it, not the rest of the line. A name that ran on to the end of a paragraph would be no shorter than the note, which rather defeats having one. The rest of the line is still part of the note.
• Any /title in a note names it, including one in a note about the command — so a note reading "see the /title command in the guide" is called "command in the guide". There is no way to have both, and a rule that guessed which you meant would be one you could not predict. Delete the command to take it back.
• The note keeps what you typed. The command stays where you wrote it, in a different colour so you can see it is an instruction — and the sentence it names is marked too, so you can tell at a glance which words the note is called by. Nothing is taken out of your text, and you can change your mind by deleting the command.
• You see the title where the note is shortened — a row in the list, a block in the calendar, a reminder, a folded card. On the wall the note reads exactly as you wrote it.
• Every note has a title, command or not. Without one it is the first sentence, and that sentence is marked in just the same way.
• One wins. A /title at the start of a line comes before one part-way through a line, even an earlier one; otherwise the one on the earliest line wins. On a single line with two part-way through it, the later of the two wins. The others stay as ordinary text, so a note cannot quietly rename itself as it grows.
• /title on its own does nothing until you finish the sentence. Until then the note keeps using its first line.
• /titles are hard is just text. The command needs a space after it, so a word that happens to start the same way is left alone.
• Capitals and indentation do not matter. Nobody types carefully while capturing a thought.

/reminder <when> Works now

Premium. Being told about this later, described in your own words.

Rather than a date picker, you say when you want to be told and it works out what you meant.

call the dentist about the crown quote /reminder thursday morning

call the dentist about the crown quote /reminder thursday morning
⏰ in 2d

• It can go anywhere — its own line, or part-way through one, which is how it usually gets typed. What you wrote before it stays in the note.
• Your device reads it. 'thursday at 8:30am', 'tomorrow morning', 'in 20 minutes', '25 Sep at 4pm', 'tonight' and the like never leave your phone and work with no signal at all.
• It shows you what it understood, straight away. The card gets the time on it — "in 2d" — so a reminder that landed on the wrong day is visible immediately rather than at the moment it fails to be useful.
• The command stays in your note. That is how you can see what it was told and change your mind: correct the phrase and the reminder moves, delete it and the reminder stops.
• Words it cannot read set nothing. "when the builders finally call back" is not a time, and rather than guess at one it leaves the note alone — you still have the note, you just do not have a reminder. Pick a time from the note's own menu instead.
• 'Morning' and 'evening' mean your hours, the ones your working day is set to. If you start at seven, 'thursday morning' is seven.
• The bell always knows, even when nothing pops up. A reminder that has come due is counted on the bell in the header whatever your device thinks of notifications, and the card carries the time either way. Nothing is lost by never allowing them — you just have to look.
• A pop-up needs your permission, once. You are asked the first time something is actually due, not on the day you sign up. Refuse and nothing is marked as delivered, so allowing it later still works.
• On an iPhone, add it to your Home Screen first. Safari only offers notifications to an app that has been added there — Share, then Add to Home Screen. Until you do, the bell and the card are how a reminder reaches you.

Reminders that reach a closed app Works now

Premium. And exactly what that costs you in privacy, which is the honest part.

A reminder used to be worked out on your device and announced there. That worked, and it had one flat limitation: a reminder falling while the app was shut everywhere announced itself when you next opened it, not at the time.

Reaching a closed app means our server has to know a reminder is due. There is no way around that and we are not going to dress it up — so here is the whole of what changes. Push is on from the start, because it carries no words; email and text wait until you turn them on.

Settings → Reminders
Reminders arrive by:
  Push   Email   Text message

a note id, a moment, the channels
and the note's TITLE, but only for
a channel that carries words

• Only push starts on, and it carries no words. With push on, we are told a note id and a moment, nothing else. Email and text, which carry a title, stay off until you turn them on. You can turn push off in Settings → Reminders.
• Push and the bell carry no title. A push wakes the app and the app reads your own library, so all we hold is a time and an opaque id. If push is all you have on, you are no more exposed than you were.
• The push itself is empty. Not encrypted — empty. Your device is woken, and then it looks at its own notes and shows you what is due. The words you see never crossed the internet. The flip side is that a device which has not synced will wake up and find nothing to say.
• It goes through your browser’s push service — Google’s, Mozilla’s or Apple’s. That is the only way to reach a closed app, and it means that company knows we woke your device and when. It does not know what for.
• Your browser asks you, once, when you turn Push on — not on the day you sign up. Refuse and the switch goes back off rather than pretending; the bell still counts every reminder that comes due.
• On an iPhone, add it to your Home Screen first. Safari will not offer notifications to a page at all until it has been installed — Share, then Add to Home Screen. Open it from there and turn Push on inside it.
• A reminder to a closed app can be up to 15 minutes late. Our server looks for reminders that have come due every 15 minutes, which is the most often the web host we use allows. A push, text or email therefore arrives within a quarter of an hour after the time you set. The bell and the card in an open app do not wait for it.
• Turn Push off and the subscription goes too, not just our note of it. Leaving one behind would mean a push service still holding a live address for a device that said it did not want one.
• Text, WhatsApp and email carry the title. A message saying “you have a reminder” is not worth sending. It is the title only — never the body, the tags, the sub-tags, a checklist's items or an attachment — and it is cut short so it fits one message. A title is still your writing, which is why it is your choice.
• A channel appears only if it can actually deliver. An option that cannot is worse than one that is missing: you would turn it on, turn the others off, and stop being reminded. WhatsApp is not offered yet because Meta requires an approved template before a business may start a conversation.
• Text messages cost credits, the same as a note texted in, and it is booked before it is sent. Email and push cost nothing. Running out stops the channel that costs money, not the reminder — the push still arrives.
• Turn a channel off and we forget. Your device sends its whole list whenever anything changes, so switching email off rebuilds every registered reminder without a title. Delete the note and the reminder goes with it.
• A milestone is not a reminder and never leaves. Our server is not told when anything is due — only when you asked to be told.

The full table of what travels where is on the Privacy page.

/milestone <when> Works now

When the thing is due, rather than when to be nudged about it.

A milestone is a deadline. It puts the note on the calendar as a bar across that day — the ◆ marks — and it is what the list view's Due column shows. It does not announce itself; that is what /reminder is for, and a note can carry both.

Ship the rewrite /milestone next friday

Ship the rewrite /milestone next friday
◆ on Friday, in the calendar

• The same words as /reminder. 'next friday', 'tomorrow', '25 Sep', '2026-12-01', 'in 2 weeks' — one way of saying when, whichever command you are giving.
• A day, not a time. A deadline is a date; the calendar draws it across the whole day, so an hour would be saying more than you meant.
• "Tomorrow" is tomorrow where you are. Your time zone is taken from this device and kept on your account, so a note you text in at one in the morning is dated in your day and not the server's. Set it by hand in Settings → The calendar and the clock if you are working to somewhere else's day. The same goes for what "morning" means: it is the hour your working day starts at.
• And it is not late until the day is. Something due today says ⏰ today all day, whatever hour it is — it turns red tomorrow morning, which is when you would call it overdue yourself. A /reminder is different: that one is a time of day, and being twenty minutes past it says so.
• It can go anywhere in the note, and what you wrote before it stays.
• Words it cannot read set nothing, rather than guessing at a date. You still have the note.
• Both at once is normal. "/milestone friday /reminder thursday morning" — due Friday, nudged Thursday.

/due <when> Works now

The same thing as /milestone, under the word you were going to type anyway.

Nobody says "milestone" when they mean a deadline, so /due sets the very same date — one field, one calendar bar, one Due column. They are two names for one thing, not two things that might disagree.

Renew the domain /due 30/09/2026

Renew the domain /due 30/09/2026
◆ on 30 September, in the calendar

• Every date the search box takes, this takes. 30/09/2026, 30Sep2026, 2026-09-30, today — the same reading as after: and on: in a search, because a date you have learnt to type once should work everywhere you can type one.
• And every phrase /reminder takes. 'next friday', 'tomorrow', 'in 2 weeks'.
• A day, not a time, and not late until the day is — exactly as /milestone, because it is the same field.
• Words it cannot read set nothing rather than guessing at a date. You still have the note, with the words in it.

/pin Works now

Keep this one at the top of the wall.

Write /pin at the start or the end of a line. The note goes to the top and stays there — the same pin as the ★ on the card, Pin to the top on its menu, and Pin on the selection bar.

• It pins; it does not unpin. The command is read again every time you edit the note, so one that toggled would flip it each time you touched it. Unpin with the ★ — that is what it is for.
• The words stay in the note, so you can see why it is up there and take it back by deleting them.
• Nothing follows it. The note it is written in is what gets pinned, so there is nothing to say after the word.

/stripes Works now

Shade every second line of this note.

Write /stripes at the start or the end of a line. Every second line of the note gets a shaded band behind it. Only the background changes: the words, tags, links and every other colour stay exactly as they are.

Line1
Line2
Line3

Line 4
/stripes

Line1
Line2Line3

Line 4/stripes

• A line is what you pressed return for. A long line that wraps over several rows on a narrow screen is still one line, with one band, so a phone and a computer shade the same lines.
• Blank lines are skipped. They get no band and are not counted, which is why Line 4 is shaded although a blank line comes before it.
• A line that holds only a command is not counted — the /stripes line itself, or one with only /pin. A line with words and a command counts.
• The word stays in the note, marked as a command, so you can see why the note is striped.
• It does not toggle. Delete the word to stop the stripes. Writing it twice is the same as once. As with /pin, the command is read again every time you edit the note, so one that toggled would flip each time you touched it.
• On a checklist, the words above the list are counted first and the items carry on the count. Ticked items count too.
• In a formatted note, each heading, list item and quote line is one line. Your formatting and highlights stay, on top of the band.
• Where you see it: on the open card, on the wall and on a board, and on a Full print. Not on a folded card, in the list or the calendar, which show the note's name, or while you are typing in the card. In the panel's Note field you see the word itself.
• It goes wherever the note goes, because it is part of the words: every device, light and dark, and notes sent from your phone or a script with /stripes in them.

/ocr Works now

The words in a photograph, added under the note.

Photograph a whiteboard, a receipt, a page of a book. Write /ocr on the note and the words come back as text underneath — searchable, taggable, and yours to edit.

• The picture stays. Reading a photograph is not the same as replacing it, and the thing you photographed is usually worth keeping beside what it said.
• The text is added, not substituted. Whatever caption you wrote stays where it is.
• It costs credits, because it costs us real money — see the plans page. The price is the same whatever the picture weighs.
• The first picture on the note is the one it reads. A note with two photographs reads the first; put the second on a note of its own.
• It waits for you. A note that arrives by message carrying /ocr is not read until you open or edit it here — it spends credits, so it does not happen while you are asleep.
• No words is an answer. A photograph of a beach comes back saying so rather than describing the beach.

/note Works now

Several notes in one message.

Write /note between two thoughts and they become two notes. Put it between six and you get six.

buy milk #PERSONAL
/note call the dentist /due tomorrow
/note book the car in

buy milk #PERSONAL
call the dentist ◆ tomorrow
book the car in

• One message, one charge. This is what it is for. A message from your phone costs credits to receive however many notes come out of it, so emptying your head on the bus costs the same as sending one thought.
• Each piece is a whole note. Its own tags, its own /title, its own /due — read exactly as if you had sent it on its own.
• Nothing is inherited. A tag in the first piece does not file the second. They are separate notes, and separate notes do not share a tag unless you write it twice.
• It works wherever you type. Sent by WhatsApp or text, pasted onto the wall, or typed straight into a card.
• A separator with nothing beside it is ignored — a stray one at the start or end, or two in a row, makes no empty notes.
• It has to start a word. example.com/note and /notes are not separators, so a link in your message survives.

/summarize Works now

The short version of a long note, or of the page it links to.

Write /summarize at the start or the end of a line. When the note is saved, a summary is added to it — of the note, or, if the note has a link in it, of what is at the other end of that link.

https://example.com/long-article
/summarize

https://example.com/long-article

The piece argues that …

And on a note you wrote yourself:

Three pages of meeting notes…
/summarize

Three pages of meeting notes…

Budget signed off, launch moved to …

• A link wins over the words. If the note contains a link, the link is what gets summarised — that is almost always why the link was saved. To summarise the note despite its link, there is nothing to do but remove the link first.
• Name a link when there are several. /summarize https://example.com/b picks that one. Otherwise the first link in the note is used, because "first" is the only rule you can predict without being told it.
• Nothing is ever replaced. The summary is added to the note, underneath what was already there. A command that ate the note it was asked to shorten would be unforgivable.
• A summary brings no link preview with it. Once a summary is on the note, an address in it is left as plain text and any /command in it is left alone — the words came from a model, not from you, and this app does not fetch pages you did not ask for. Edit the note yourself and it is an ordinary note again. The same is true of /ocr and of anything the assistant writes; Link previews says it once.
• A short note is refused, and says so. Summarising two sentences into two sentences is a round trip, not a service.
• A link that cannot be read is an error, not a fallback. If the page is unreachable you are told. The note is not quietly summarised instead — you asked about the link.
• The line goes even if the summary does not arrive. /summarize is an instruction about the note rather than part of it, so it is taken out straight away. Your note is untouched otherwise.
• The first one wins. As with every command. A second /summarize further down stays as ordinary text.
• /summarizes the meeting is just text. The command has to be the whole word, at the start or the end of a line.

/checklist Works now

Turns lines into a list of things to tick off — from the command's line to the end of the note, one line, one item.

Write the list first and label it afterwards, or start with the command and keep typing. Both end in the same place:

/checklist Milk
Bread
Eggs

☐ Milk
☐ Bread
☐ Eggs

Written lower down, under some words of your own, it starts the list on its own line. The lines above it stay ordinary text, in the order you typed them:

This is a test.
/checklist Line1
Line2

This is a test.
☐ Line1
☐ Line2

• One line, one item. Where you pressed return is where the list divides.
• The list starts on the command's line and runs to the end of the note. The lines above it stay your words, shown as ordinary text above the boxes. A blank line does not end the list, and neither does another command.
• With nothing after the command, the whole note is the list. /checklist alone on the last line turns every line above it into an item — that is the write-the-list-first way. On the first line it does the same.
• What you write after the command is the first item, not something thrown away. Typing / offers the command and leaves the cursor after it, so carrying straight on is the natural thing to do. Words before it on the same line are the first item too.
• A space goes after the command. /checklistLine1 is not the command and stays as text, the same as /titles is not /title. Choosing the command from the menu puts the space in for you.
• The command itself disappears. It is an instruction about the note, not part of it.
• Blank lines are not items.
• A line below that holds only another command, such as /pin, is not an item. It stays with your words above the list and works as it does anywhere in a note.
• A second /checklist does not start a second list. The command goes, and the rest of its line is an item.
• The Checklist button does the same for the whole note. Opening a note and switching it to a checklist in the panel gives what /checklist on the first line gives.

A checklist can go back to being an ordinary note the same way, and the items become lines again, after the words above them, in the order they were in.

/object Works now

Add a mind map or a data model to this note.

Write /object mindmap on a line and save the note. The words are taken out, and a new mind map opens full screen, kept with the note. /object datamodel does the same with a new data model. Type /object and the command list offers the kinds there are. Typing /mind finds the mind map too, and /data, /model or /erd the data model. See Mind maps and Data models.

• It acts once, when you write it. Only the words you typed in this edit count. A note that already said /object mindmap, such as one you emailed in, never grows a map when you edit its other words.

#tag, mid-sentence Works now

A hash and a word anywhere in a note files the note there.

Write the tag where it belongs in the sentence. It stays in your words and is coloured to show it is a tag, and the note is filed under it.

ring #home about the boiler

ring #home about the boiler
filed under #HOME

• Your words are not edited. An earlier version took the tag out of the sentence, on the grounds that it appears as a chip anyway. That was wrong: the tag is part of what you wrote, and removing it rewrites your note. It is coloured where it sits instead.
• Type # and the tags you have appear, each in its own colour. Arrows to move, Enter or Tab to take one, Esc to dismiss the list. Typing past the end of it is how a new tag gets made.
• A sub-tag brings its parent. #health/pharmacy files under #HEALTH as well, so filtering by the parent still finds it.
• The hash has to begin a word. C# is a language, #ff8800 is a colour, and the #section on the end of a web address is part of the address. None of them become tags — a rule loose enough to file those would make the character frightening to type.
• It works on a note you already have. Edit it, add #urgent, and it is filed when you finish.

#tag Works now

How a note is filed, and how you find it again.

Write a hash and a word anywhere in the note. Tags are the spine of the app — the rail down the side, every filter, every view — so most notes want at least one.

the boiler serial plate — the engineer
wanted this before quoting #home #action

the boiler serial plate — the engineer wanted this before quoting #home #action
filed under #HOME #ACTION

• Case does not matter. #dev, #Dev and #DEV are one tag, not three.
• A tag ends at the first space or punctuation mark, so #next week files under #NEXT. Dashes and underscores do not end it: write #next-week to keep the two words together. A slash starts a sub-tag, and accented letters are letters: #café files under #CAFÉ.
• Letters, numbers, dashes and underscores. Punctuation is refused, because a tag has to survive being read back out of a sentence.
• A tag goes on a note once. Writing it twice does not file it twice.
• Up to 32 characters, and up to 40 tags on one note.

#parent/child Works now

A sub-tag, for when one tag is getting crowded.

Once #HEALTH has two hundred notes in it, it has stopped being useful. A sub-tag narrows it without starting again.

You can also make one from the + tag button on a note: type /joe — just the slash and the name — and it becomes a sub-tag of the tag the note already carries. If the note has two tags it asks which one rather than choosing for you; if it has none it says so, because a sub-tag needs a parent.

repeat prescription runs out on the 14th
#health/pharmacy

ring them about the quote
#PALL/joe/matt

repeat prescription runs out on the 14th
filed under #HEALTH/pharmacy

ring them about the quote
filed under #PALL/joe and #PALL/matt

• A sub-tag brings its parent with it. Filtering by #HEALTH finds this note, which is the whole point of having filed it under a child of it.
• Several at once: #PALL/joe/matt. That is #PALL/joe AND #PALL/matt — a parent, then each of its children. Shorthand for writing the parent twice, and as many children as you like.
• Still one level deep. It is not a deeper hierarchy: #PALL/joe/matt is two sub-tags of #PALL, not a matt inside a joe. Two levels is a filing system; three is a filing problem.
• The same child twice is one. #PALL/joe/joe is one sub-tag, for the same reason #DEV #DEV is one tag.
• The same name under two parents is two sub-tags. #LMG/test and #POLL/test are separate: each files its own notes, and one note can carry both.
• A child written without its parent is decided once. If you have a main tag called #TEST, #test is that. If not, and a main tag already has a test sub-tag, the note goes under the first such main tag in the rail's order — pinned first, then A to Z — without asking, and the chip shows which. With neither, it becomes a new main tag. A note that arrives by email, text or a script is decided the same way. The answer stays on the note: changing your tags later re-files nothing. Write the parent when it matters: #POLL/test.
• Take the last sub-tag off and the parent goes too, so a note never keeps turning up under a tag with nothing on it to explain why.

They stack up in the rail. Every tag you have used is listed down the side, sub-tags folded under their parents. Click one to show only the notes that carry it; click it again to hide those notes instead; a third click lets go. It never changes the view itself — that stays however you set it up.

Main, pinned, and everything else Works now

Three words for three different things — and none of them change a single note.

As far as a note is concerned there is only one kind of tag. These three words are about the filter rail along the top — which tags it shows, and in what order. Promoting, demoting or pinning never adds, removes or alters a tag on anything.

a main tag
pinned
an ordinary tag
a sub-tag

always in the rail
first in the rail
in the rail behind “+N more”
sits under a main tag

• A main tag is always drawn, and so is a pinned one, even when no note on screen carries it. Everything else folds away behind +N more until you ask for it. That is the whole difference — it is about whether a tag is in front of you, not where.
• Pinned is about order. A pinned tag comes first, before every other tag, main or not. Pinning is the only thing that moves a tag up the list; a dozen main tags still leaves two or three you reach for daily, and alphabetical order has no opinion about which.
• Everything else is A to Z. Pinned tags first in alphabetical order, then all the rest in alphabetical order, mains and ordinary tags together. Being a main tag is not a place in the queue.
• The rail and the Main tags panel read the same way round. They are the same list, so a tag is where you last saw it in both.
• A sub-tag is a different question entirely. It is about meaning — #HEALTH/pharmacy belongs under #HEALTH and brings it along — not about the rail. See #parent/child above.

All of it lives in Main tags, from the button above the wall: promote, demote, pin, rename, recolour, add sub-tags, and remove a tag.

Removing a tag

Every tag and every sub-tag in Main tags has one button, Remove from all notes and delete. It asks first, and the question says how many notes it will touch.

• The tag comes off every note that carries it, archived notes and notes in the trash included, and then it is gone from your list of tags. If you wrote it in the words, as #PALL, it is taken out of the words too, or it would be read straight back.
• The notes themselves are kept. Only the tag goes.
• A main tag takes its sub-tags with it, and the question says how many. Removing one sub-tag leaves its parent and the other sub-tags alone.

Renaming one

Click its name and type over it. That works for a tag and for a sub-tag — click #PALL on the row, or click #mike on the chips underneath it — and it is the same gesture either way.

• It follows the name everywhere. Every note carrying it, the rail, the colour you chose, the context bar, and what new notes start with. A rename that only changed the list would quietly untag every note that had it.
• Including the word in the note. If you wrote #PALL/mike in the sentence, the sentence now says #PALL/michael. It has to: the tag in your words is read back out of them every time the note is drawn, so a rename that left the word alone would put the old tag straight back on.
• Renaming onto a tag you already have merges them. Rename #WORK to #ADMIN when #ADMIN exists and the notes join it. That is what typing the name of a tag you already have means.
• A sub-tag is renamed under its own parent. Renaming #LMG/test to #LMG/trial leaves #POLL/test alone, because it is a different tag. Renaming it onto a sub-tag #LMG already has merges the two. Renaming a main tag takes its sub-tags with it: #LMG/test becomes #WORK/test.
• A sub-tag may have the same name as a main tag. #LMG/test and a main tag #TEST are two different tags, and the full path tells them apart. Written on its own, #test means the main tag, as #parent/child explains.
• A tag is UPPER and a sub-tag is lower, however you type it — the same rule as everywhere else.

Moving a sub-tag to another main tag

Renaming never changes a sub-tag's parent; moving does. Press Move to… on the sub-tag and choose the main tag.

• It moves on every note, archived notes and notes in the trash included, words included: #LMG/test written in a sentence becomes #POLL/test. An Undo puts it back.
• If the other main tag already has that sub-tag, it asks first. It says how many notes are under each, and the button becomes Merge into #POLL/test. Press it again to merge the two into one.
• The old main tag stays on a note while the note still has another of its sub-tags, or has #LMG written on its own in the words.

Merge tags Works now

For when a typo became a tag, or two tags mean the same thing.

Press ⇄ Merge tags above the tag chips. In Merge this tag, type a few letters and pick the tag that should go. In Into this one, pick the tag to keep. Either can be a main tag or a sub-tag.

Type lmg/ to list only #LMG’s sub-tags, and lmg/te to narrow them. A tag you have chosen always shows its full path, so #LMG/test cannot be mistaken for #POLL/test. Nothing is created here: to give a tag a new name, rename it in Main tags.

From Main tags there is a shorter way: press Merge into… on a sub-tag. The sheet opens with that sub-tag already chosen and the other sub-tags of its main tag listed first.

Before you press Merge, the sheet says what will happen and to how many notes, archived notes and notes in the trash included. Merge acts on one press, and the message that follows has an Undo.

#LMG/test into #LMG/tests
#LMG/test into #POLL/team
#LMG/test into #POLL/test
#LMG/test into #FOO
#FOO into #LMG/a
#FOO into #BAR

lose #LMG/test, gain #LMG/tests
gain #POLL/team, and with it #POLL
join #POLL/test, as Move to… does
gain #FOO
gain #LMG/a, and with it #LMG
gain #BAR; #FOO's sub-tags move under it

• Every note that carries the tag changes, archived notes and notes in the trash included. A note in the trash stays in the trash, but it no longer holds the merged tag, so restoring it later does not bring that tag back.
• A note that carried both ends with the kept tag once. Nothing else on the note changes.
• The words change too. If you wrote #LMG/test in the sentence, it now says #LMG/tests. It has to, for the same reason as a rename: the tag in your words is read back out of them, so a merge that left the word alone would undo itself.
• A sub-tag still brings its main tag. Merge into #POLL/team and the notes are under #POLL too. The old main tag stays on a note only while the note still has another of its sub-tags, or has #LMG written on its own in the words — the same rule as Move to…. Merging #LMG/test into #LMG itself takes the sub-tag off and leaves #LMG.
• Merging two main tags brings their sub-tags along. #FOO/x becomes #BAR/x, on every note. Where #BAR already has an x, the two become one. The sheet lists which sub-tags move and which join one that is already there.
• A main tag that has sub-tags cannot go into a sub-tag, because a sub-tag cannot have sub-tags of its own. The sheet says so and Merge stays unavailable. Merge it into a main tag instead, or move or merge its sub-tags first.
• The tag you keep keeps its name, colour, pin and place. The merged tag’s colour and pin go with it.
• Everything that named the merged tag follows. A saved view that filtered on it now filters on the kept tag, and so do your context tags and Tags on a new note. Board columns are not tags, so no column changes.
• A merge never sends more to the assistant. If either tag was on your Never send these tags list, the tag you keep is on it afterwards.
• Undo puts everything back: each note’s tags and words, the merged tag itself, colours, pins, views and context tags. Undo is the button on the message; Ctrl+Z does not undo a merge.
• Your tag limit never stops a merge. A merge can only lower the number of tags you have.
• Your other devices follow, as each changed note syncs. A note written before the merge that arrives afterwards (by email, by text, or from a device that was offline) and has #LMG/test in it makes that tag again, as it would after a rename. Merge it again.

The star Works now

Keeping something where you can see it.

Open a note and press the star in the top corner. It moves to the front of every view, and its card gets a coloured edge so you can pick it out.

• It comes first whatever the view is sorted by, and whichever direction — a pin that sank to the bottom in ascending order would be doing the opposite of what you asked.
• Several pinned notes keep their normal order among themselves.
• A note you are making stays above them until you move on: while you type it, tag it or set its reminder it stays first, and it takes its place once you start another note, open a different one, or change the view.
• A filter still applies. Pinning keeps a note at the top of the views it appears in; it does not force it into views that exclude it.

Choosing several Works now

Tagging thirty notes without opening thirty notes.

Every card has a small checkbox at the top right. Tick one and the rest appear; after that a tap picks another note instead of opening it. A bar appears with what you can do to the lot: Select all shown, Tags, Milestone, Collapse, Pin and Delete — plus Move to in the list view, which files them under one of its categories.

Priority is not among them, and is not on the note either: it belongs to the list view, where it is a column you set a row at a time.

• The same rules apply as to one note. A tag that would be refused on a single note is refused on all of them — nothing gets in by the back door because you were in a hurry.
• Some already being done is not a failure. Tag thirty notes when four already have it and it tags the twenty-six, leaving the four alone. The sheet says so before you start.
• Your picks stay after an action, so you can tag them and then set a milestone on the same selection without starting again. Done clears them; deleting clears them too, because they are gone.

Opening a note Works now

How long you hold decides what you get.

A tap does nothing at all, on purpose — a wall of notes is something you scroll past, and a list where brushing a card opens it is a list you cannot read. Holding is what opens one, and how long you hold chooses how.

a moment
about a second
about three seconds

nothing — it was a scroll
edit it where it sits
the full editor, in the side panel

• Editing in place is for a word or a line. The card becomes editable where it is. Enter makes a new line, because a note is prose; Ctrl/⌘+Enter finishes, and so does tapping away.
• A double click does what a one-second hold does, for a mouse. Neither starts from a button or a link. A hold may start on a file or its picture on the card, and edits the note, so a note that is only a file is edited the same way; a tap or a double click there opens the file.
• A note with formatting always opens the side panel. A quick fix in place would show only the words and drop the formatting without a word, so a hold, a double click or F2 on a formatted note, in any view, goes straight to the full editor. See Formatting a note.
• A checklist is edited item by item. Hold the line you want and the caret lands in that line. Enter starts the next item, Backspace at the start of an empty one removes it, and Ctrl/⌘+Enter finishes. The tick boxes are not editable — they are state, not words.
• Esc puts it back as it was before you started. That is the only way to abandon an edit — tapping away keeps what you wrote.
• The side panel is for the rest — tags, a reminder, a file, the things that are not the words themselves.
• A link in a note is still a link. Tapping it opens the page rather than the note.

Viewing a note large Works now

A long note, or one with a table, read at the size of the window.

A card is narrow, so a long note runs far down it and a table's words break letter by letter. View large shows the whole note over the page, the way a picture opens: as wide as the window allows, and the whole screen on a phone.

• Three ways to open it. The ⤢ button at the top of a card, beside ⋮, which shows when you point at the card, click or tap it, or reach it with Tab. View large in the note's ⋮ menu, in every view — on the Calendar, in the menu of the note's piece. And Space on a note you have clicked.
• Tables get their full width. A table wider than the window scrolls sideways in its own box; the words around it never do.
• It is for reading. You cannot type in it, but everything that works on the card works here: tick an item, open a link, a picture, a file or a mind map. To change the words, press Edit: it closes and opens the note in the side panel.
• Close it with Close, Esc or a click outside it. Selecting words and letting go outside it does not close it. If a picture is open on top, Esc closes the picture first.
• If the note goes while you read it — moved to the Trash or deleted, here or on another device — it closes and says so.

Formatting a note Works now

Bold, lists, headings, tables and links — in the note and on its card.

A note is plain words until you say otherwise. Open it in the side panel (see Opening a note) and a toolbar sits above the words. Press one of its buttons, or paste something that already has formatting, and the note becomes a formatted note in place, with your cursor where it was.

B       Bold (Ctrl+B)
I       Italic (Ctrl+I)
H       Heading
•       Bulleted list (Ctrl+Shift+8)
1.      Numbered list (Ctrl+Shift+7)
▦       Table
🔗      Link

- or * then a space   a bulleted list
1. then a space      a numbered list
#, ## or ### and a space   a heading
**word** and a space   bold
_word_ and a space     italic
| a | b | then Enter     a table

• The card shows it too. A formatted note is drawn formatted on the wall and on the board, so you do not have to open it to see the table. The card's usual rules still apply: folding, the height limit and its more, colours, pin and tags. In the list, the calendar and a folded card you see the words and the title, as before.
• Pasting keeps the formatting. Copy from Word, a web page or another note and paste: into the open note, into a new note on the wall, or into the composer. Headings, bold, italic, lists, tables and links come across. Colours, fonts and sizes do not, and a picture becomes its description and address as words. A plain-text paste stays plain.
• Tables. The table button adds one, and a menu offers Add row above, Add row below, Add column left, Add column right, Delete row, Delete column and Delete table. A table with merged cells, such as one pasted from a document, keeps its shape: its rows and columns cannot be changed here. On a card a table is fitted to the card's width; in the open note it scrolls sideways if it is wider.
• Links. The link button asks for an address that starts with https:// or http://. An address in a formatted note previews like any other.
• Search, tags, commands and the assistant read the words. Formatting never changes what a search finds, what a #tag or /due does, or what is sent when you ask the assistant. The words are always kept beside the formatting.
• If the words were changed somewhere that cannot format — a script through the API, a command, an older copy of the app — the note shows its plain words, with a line saying why, rather than formatting that no longer fits. Editing it removes the old formatting; pinning or tagging it does not.
• Checklists are not formatted, and a note that arrives by text, WhatsApp, email or the API arrives as plain words.
• It travels with the note to your other devices and into your backups.

Deleting a note Works now

Swipe left, and a way back afterwards.

Swipe a card to the left and a Delete button appears on it. Press it and the note goes. Whether you are asked first, and whether it goes to the trash or straight out, are both yours to set — see Settings.

• The trash is on by default. A deleted note waits there instead of ending. Open it from the menu at the top right, tick what you want and put it back.
• Nothing empties the trash on a timer. It is on your own device and it stays until you empty it — a note you can still get back is the entire point, and a countdown running quietly against that would make the promise worth less.
• Turn the trash off and delete means delete. Said plainly here because it cannot be undone afterwards.
• Esc cancels the swipe, and so does tapping anywhere else.

Archiving a note Works now

Swipe right: put away, not deleted.

Swipe a card to the right and an Archive button appears on it. Press it and the note leaves the wall without going to the trash. Archived notes are hidden by default, so a note put away really does go away.

• The menu does the same. The ⋮ on a note has Archive (E once the note is clicked), and, for a note already archived, Bring back from the archive. Swiping an archived card right shows Unarchive.
• The Archived chip is how you get at them. At rest it reads Archived hidden. Each press moves it on: Archived shown puts them back among the rest, and Archived only shows nothing else.
• The swipe works the first time. Left deletes, right archives, on the wall, the board, the list and a saved view alike. Press and pause before you swipe and it still counts as a swipe. Esc cancels it.

Add to my calendar Works now

A note with a time on it, in the calendar you actually use.

In the Calendar view, open a note and press the ⋮ on it. Add to my calendar saves an .ics file — the format Outlook, Apple Calendar and Google Calendar all open.

• A file, not a connection. Putting something in your real calendar would mean holding a token for that calendar, which is the one thing this app is built not to do. An .ics is the interchange format that exists so software does not have to — and it works with a work Exchange account that would never hand out a token anyway.
• Adding it twice updates it. The file carries the note's own id, so your calendar replaces the appointment rather than putting a second one beside it.
• A milestone becomes a whole day, and a reminder becomes an appointment of thirty minutes — or whatever length you gave it. A repeat comes with it.

Views Works now

The same notes, arranged for what you are doing.

The wall is one view. The tabs along the top are others — a board with columns, a week, whatever you make — and each one remembers its own layout, sort order, width and which tags it is filtering by.

• A view is a way of looking, not a folder. A note is not "in" a view. Nothing is moved or copied, and deleting a view never touches a note.
• The tag chips filter what you see. Any shows notes with any of the tags you have picked; All shows only the ones carrying all of them. Filtering by a sub-tag shows the notes filed under that sub-tag of that parent; filtering by the parent also finds notes filed under any of its sub-tags.
• A note you make in a filtered view starts with that view's tags. Make a note, add a card, drop a file or paste a picture while tags are picked in the view, and the new note is filed under those tags, so it stays in the view instead of vanishing from it; a short message names them, with Undo, and the tags come off like any other tag. These tags come before your context tags and the tags set for new notes. Pinned, Untagged and the search box add none, and notes that arrive by email, text or sync are not touched.
• There are four kinds of view. The tabs start with a Wall (cards), a List (rows with columns you choose), a Board (columns, with cards you drag between them) and a Calendar (notes placed by a date, with Day, Week and Month to change the stretch and Today to come back). + New view makes another of any kind. Keeping views of your own is a Premium feature; see Free.
• Save, Save as a new view, Delete view. The gear on the tab opens that view's settings. Save keeps your changes in it. Save as a new view copies everything the view is set to: its kind, filter, sort and display, and its columns or categories with the notes in them. Only the name is new. Delete view is at the bottom left, appears for a view you made, and never touches a note.
• Drop a card anywhere in a column. On a card, between two cards or on the empty space below them, the whole column takes it. You do not have to aim at a particular spot.
• Dragging a card between columns is a real change, and it belongs to that board. It survives a reload and travels to your other devices — but a second board over the same notes keeps its own arrangement, so moving a card here does not move it there.
• Each board has its own columns, and each list its own categories. That is the point of making a second one: the same notes, sorted into stages for a project on one board and into something else entirely on another. Renaming or deleting a column on one board leaves every other board alone.
• A column has its own menu. The ⋮ beside a column's name will rename it, give it a colour, move it left or right, fold it out of the way, or delete it. Deleting a column never deletes the notes in it — they go back to Unassigned.
• Drag a column to put it somewhere else, the way you drag a view along the view bar. Hold its header — anywhere on it, or anywhere on a folded column's strip — and move sideways: the column moves between its neighbours as you go and stays where you let go. With a finger or a pen, rest on the header for half a second, then move; moving sooner scrolls the board as it always has. Near the board's edge it scrolls along with you. Press Escape before letting go and the column goes back where it was. A click that does not move still opens the menu or folds the column. Unassigned moves like any other column; + Add column always stays last.
• From the keyboard, Move left and Move right in the column's menu do the same, as do Alt+← and Alt+→ on its ⋮. The new place is announced, for example “Moved left, 2 of 5”.
• Moving a column changes only the order of that board. Other boards and your lists keep theirs, and the new order is the same after a reload and on your other devices. No note changes column or place, and none is edited. A filter or a search hides cards, never columns, so a column you move in a filtered board moves for the whole board, with the cards the filter hid. To undo a move, drag it back or use Move left and Move right.
• What a column looks like belongs to the view, not to this device. Its name, its colour, its place in the row and whether you folded it are saved with the account, so a board you arranged on one device opens arranged on the next — and a column you add on your phone and one you add on your laptop both survive.
• A colour you pick is used as you picked it. The colours a column starts with follow the theme and stay legible in light and dark; choosing your own says what you want, and it is kept exactly — the same bargain a tag's colour makes.

Print a view Works now

Paper or a PDF of the notes you are looking at.

Open the view you want, and press Print on the strip beside Tags. On a narrow screen the strip has no room for it, so use the Print this view row in the view's settings (the gear) instead. On a computer you can also press Ctrl+P (Cmd+P on a Mac). A sheet opens first, so you choose what goes on the paper before your browser's own print dialog appears.

• What is printed is what the view shows. The tags you are filtering by, a search, and whether archived notes are included all apply, and the top of the page says so. Every note in the view is printed, not only the ones that have scrolled into sight. The sheet tells you the count and about how many pages that is before you print.
• Full or Compact. Full notes prints all the text, every checklist item and the dates. Compact prints the title, tags and dates only, about four times as many to a page. Tags, the due date and reminder, the created date and link titles can each be left out. The sheet remembers what you chose, on your account, and opens the same way next time.
• Pictures small, or off. In Full, a note's pictures are printed small, and other files are listed by name. A picture shown in the note's words is printed at its place there, and the others after the words; none is printed twice. Untick Show pictures, small to leave the pictures out: a shown picture's place is left out too, and its file is listed by name with the others. Compact never prints them.
• A big view takes a while. From about 250 notes the sheet says your browser needs a moment to get the view ready, and to keep the tab open. Above 500 notes, or about 40 pages, in Full, it says that is a lot of paper and offers Use Compact, which shows the pages Compact would use.
• A Board prints sideways. The paper is landscape and the columns sit side by side, in the board's own order. A board with more than four columns gives each column its own section. Each column's name starts with a filled circle in the column's colour.
• A Calendar prints the month, the week or the day you are looking at, as a grid on landscape paper, or as an Agenda, day by day, on portrait paper with full notes. Choose which under Show as. A grid shows only the title and time of each note.
• It always prints black on white. The dark theme is for the screen. Paper gets dark text on white, and your tag colours, made darker where they would be too pale to see.
• It works offline. The notes are already on this device, and printing asks the server for nothing.
• Nothing is sent, and no credits are used. Printing happens in your browser on this device. The app sends nothing and uses no credits.
• The copy is not sealed. A paper or PDF copy is not encrypted, and anyone who can reach the paper, the file or the print queue can read it.
• Page numbers and the footer appear in Chrome and Edge only. There the page says Page 1 of 3 at the bottom right, and the view's name, the day and the time you printed at the bottom left, and nothing else is printed at the top or the bottom: the browser's own Headers and footers option (its date and time, the page title and the address) has no effect on the paper, whether it is ticked or not. Firefox and Safari cannot draw our footer or the page numbers, and they keep their own Headers and footers option: if it is ticked, they print their date, the page title and the address on the paper, and you switch it off in the print dialog.
• Ctrl+P prints the view only when it will not get in the way. While you are writing or editing a note, with another sheet or menu open, or in the assistant, it stays the browser's own. The browser's own Print (the File menu, or a phone's share sheet) prints a single page that points back to Print this view, so a note or your twelve words are never printed by accident.

Keep the paper safe. A paper or PDF copy is not encrypted, and anyone who can reach the paper, the file or the print queue can read it. On a shared computer, collect the paper and delete the saved file. A saved file is named after the view.

Printing your twelve words is separate. The Print button on the recovery phrase sheet opens a small window that holds the twelve words and nothing else, and prints that. The window is emptied and closed when printing ends, or after a minute if your browser never says it has. If your browser blocks the window, allow pop-ups for this site, or copy the words instead. Save as PDF puts the twelve words in a file on this device, named Private page: delete the file once the paper is safe. See Twelve words.

Tags and colour Works now

Eighteen colours, or one of your own.

Adding a tag lets you choose its colour from the palette, or pick any colour at all. A colour you choose is kept for that tag and is not reshuffled when other tags are added later.

• A palette colour follows the theme. It is remembered as a colour name, so it stays legible in both light and dark. A colour you pick yourself is kept exactly as chosen, in both.
• Tags you never colour are given one from the palette, in turn, so the wall is readable before you have decided anything.
• A tag you write in the words has its tag’s colour wherever you write it — in the composer, on a card you are editing, in the side panel — so the same #tag does not look different from one door to the next.

Tags on a new note Works now

What a note starts out filed under, before you have said anything.

Nothing, unless you choose otherwise. Press + choose and pick the tags every new note should be born with — a note you write on the way to somewhere is often the same kind of note every time, and picking the same tag by hand each morning is a small tax on the thing the app exists to make cheap.

• These are on top of anything else. A #tag you type in the note itself still files it there as well.
• They apply to every way a note is made — the capture box, N or +, pasting onto the wall, a card started on a board.
• The × takes one off again, and only affects notes made from then on. Notes already filed stay where they are.
• A sub-tag brings its parent, exactly as it does on a note, so filtering by the parent still finds what you filed.
• Context tags overrule this while they are set — see below. This is what you want always; that is what you want this afternoon.

Context tags Works now

The same thing, for this afternoon rather than for ever.

Above the tag chips, beside Merge tags, there is ✎ Context tags. Put #WORK in it and every note you write from then on starts filed under #WORK — until you clear it with the ×.

It is for a stretch of work. You sit down to deal with the roof, set #HOME/roof, and the next hour of notes files itself.

• It replaces the setting above, it does not add to it. If you have both, the context wins — you went to the wall and said "these, now", which is a clearer answer than a preference set once in a drawer. Clear it and the setting comes back.
• It is a helper, not a rule. The tags arrive as ordinary chips on the new note, so you can take one off with its × before you save, exactly like a tag you added yourself.
• You see it before you commit. The chips are on the blank card the moment it opens, so a context you had forgotten is in front of you rather than a surprise a week later.
• Sub-tags work. #HOME/dentist files under both, the same as typing it.
• Only notes you write here. A note that arrives — texted, sent by WhatsApp, posted by a script, or synced from another device — never picks these up. You did not set a context for those, and a note sent from a bus was not written at this desk.
• It survives a reload, and belongs to this browser and this account. Signing in as somebody else does not inherit it.

On every plan.

Link previews Works now

How much of a preview card to show.

Two settings decide how many lines a preview gives the page title and the description beneath it — 2 and 4 unless you change them. The wall re-flows as you change the number, so you can see what you are choosing.

• Not every site offers a picture. Some publish one and some do not, and a few refuse to answer at all. A preview with no thumbnail is usually the site's choice, not a fault.
• A link you write previews on its own. Type or paste an address and a card with the page’s title, a line or two and, where the site offers one, a picture appears under the note a moment later. Our server opens the page to make it, so the site normally sees our server and not you. A page that cannot be read gets a card that says why, such as No preview. The site did not answer.; the note is untouched.
• A preview is kept until no note holds its link. It stays on your device, with its picture, and goes when the last note holding that link is deleted for good or emptied from the Trash. It is not looked up again unless you ask: Refresh link preview in the note's ⋮ menu does it. If the site fails then, you keep the old preview, and the app says The preview could not be refreshed just now. The old one is kept.
• A preview that failed for a passing reason tries again by itself. Its card ends It tries again later. and has Try again to do it now; if that fails too, the card starts No preview. Tried again just now: and says why. A card without those words, such as a page that was not found or a site that wants a sign-in first, has no Try again: waiting would not change it. If this device asked for many previews in a few minutes, the rest wait with Too many links were looked up at once. It tries again later. and arrive by themselves.
• A link in an emailed note previews on its own only if the note is from you. An address in an email was chosen by whoever wrote the email, and a site can tell when its address was opened. So a note you email yourself from an address you have linked and verified gets the preview of its first link as soon as the note is shown, without a press. Every other emailed note, such as a forwarded message, one that quotes another message, one from an address you have not verified, or one you brought in from Email waiting to arrive, shows Show preview instead, and nothing is fetched until you press it. Either way our server opens the link, not your device. Each device makes at most 100 of these automatic previews an hour; past that, the note waits for your press too. Press it and the note previews like any other from then on; so does editing the note yourself. Notes you type, paste, text in or send from your own script are not held back.
• You can make every emailed link wait. In Settings, set Preview links in my emailed notes automatically to When I ask, and a note emailed from your own addresses also waits for Show preview. Automatically is the default. Neither choice ever makes a forwarded or unverified note preview on its own.
• A link the assistant wrote gets no preview. If the assistant writes a note, rewrites one or adds to one — or if /summarize or /ocr puts words on a note — any address in those words stays plain text: no preview card, and no /command in them runs either.
  Why: fetching a page is something this app does for you, and it will not do it because of an address you did not write. The assistant reads your notes to answer you, and a note can hold anything, including an email somebody else sent you.
  Edit the note yourself and it is an ordinary note again, preview and all.
• You can have your own words back, later. When the assistant rewrites a note or adds to one, the note keeps what you had written before. Undo in the answer panel puts it back straight away — and if you have closed the panel, reloaded, or picked up your other device, the note’s ⋯ menu offers Undo the assistant’s rewrite until you write on that note yourself.
  Why: the assistant does what you asked and then you decide, so taking it back has to work for longer than one screen. It is your writing it gives back, not the assistant’s earlier attempt: if it rewrote the same note twice, you still get what was there before it started.
  Once you write the note, the offer goes — at that point putting a copy back would throw your own words away.

Deleting, and the trash Works now

How careful you want it to be.

• Delete at once, by default. A confirmation exists to stop something irreversible happening by accident — and with the trash on, deleting is not irreversible. Asking anyway would be a tap that guards nothing. Switch it to Ask first if you would rather.
• Turning the trash off brings the question back. The two settings are a pair: without the trash, one press really would end a note. You can still have neither — turn the trash off, then choose Delete at once again — but not by accident.
• Trash on or off. On — the default — a deleted note waits where you can get it back. Off, it is gone when you say so.

Appearance Works now

System, Light or Dark — and it stays chosen.

System follows whatever the device is doing, including when the device changes its mind at sunset. Light and Dark override it. The button in the header flips between the last two, because that is all one button can mean; System is a third answer and lives here.

• It stays on this device, which is the point of it. A laptop in a bright room and a phone at night want different answers, so this one is not carried between them.

Opens on Works now

The wall every time, or wherever you left off.

Wall — the default — always opens on the wall. Last used opens whichever view you were in when you last closed the tab.

• It remembers the view itself, not its name, so renaming it does not lose it. A view you have since deleted is not an error; it opens the wall.

Time zone Works now

What “tomorrow morning” means.

It is taken from this device and kept on your account, so a note that arrives by text message while the app is shut is still dated in your day. If you are working to somewhere else's day, choose that zone here, in Settings → The calendar and the clock.

Copies to keep Works now

How many older backups sit beside the newest one in your Drive.

Three unless you change it, and anywhere from one to ten. The newest always carries the same name — it has your account in it, so two accounts sharing one Drive never find each other's — and that name never changes, so there is one obvious file to restore from. The ones before it are .1, .2 and so on.

In Google Drive the folder is called MyNotesHelper. In OneDrive it is the app folder OneDrive gives this app. Either way it is made for you the first time a backup is written.

MyNotesHelper/
  mynoteshelper-<account>-backup.json
  mynoteshelper-<account>-backup.1.json
  mynoteshelper-<account>-backup.2.json
  mynoteshelper-<account>-file-<id>.bin

the one just written
the one before it
the one before that
one attachment, one file

• Nothing accumulates. When a new copy arrives the oldest is deleted, so the space used stops at the number you chose rather than growing by a few megabytes every time you back up.
• The new file is written and read back BEFORE anything is moved. A connection lost halfway through leaves every copy you already had exactly where it was — that is the whole reason for keeping more than one.
• Lowering the number tidies up. Copies that no longer fit are removed on the next backup, rather than sitting in your Drive where nothing in the app would ever mention them again.
• One is a real answer, and means what this did before: replace the file, keep no history.
• A downloaded backup is different — it is named with the date, because it lands in a folder full of other things and has to be told apart from them. Nothing deletes those; they are yours.

Switching the AI features on and off Works now

Off until you say otherwise, and off again in one press.

The AI features start switched off. The first time you open the assistant, or use /summarize or /ocr, a short explanation appears where you are: what is sent, who it goes to, how long they keep it, and what it costs. Nothing has been sent at that point. It has two buttons of the same size, Reject and Accept.

Press Accept and the thing you asked for happens — you do not have to ask again. Press Reject and the assistant is not shown anywhere in the app, on any of your devices, and nothing is sent; your notes work as before. Escape closes the explanation without choosing, and it is shown again next time.

• One way back. After Reject, Settings → The assistant keeps one switch. Turning it on shows the same explanation, and it switches on only when you press Accept.
• It belongs to your account, not to this device. Switch it on once and every device you sign in on already knows. The same is true of switching it off.
• Switching off is one press, and asks nothing. Under the Ask box, press Switch off; or use Settings → The assistant. The assistant leaves the app at once, and nothing is sent afterwards, from any device.
• Settings says when you agreed. Settings → The assistant shows the date and time in your own time zone.
• If the explanation changes, you are asked again, and shown only the part that is different. Nothing is sent in between.
• On a new device there is nothing to do. Signing in is enough; what you agreed to travels with the account.
• A script with a write-API key can neither switch it on nor use it. Agreeing is something a person does.

Recovery phrase Works now

Where the twelve words live, and what to do without them.

Settings will not show you your phrase, because it is not kept anywhere. What it will do is give the same key a new phrase, which is the answer if the old one is lost. See Your data. To put the twelve words on paper, see Print a view.

Attach a file Works now

A photograph, a receipt, a PDF, kept with the note.

Open a note's ⋮ menu and choose Attach a file…. Several at once is fine.

Or drag the file in, and where you drop it is the instruction:

• Onto empty space — a new note is made, with the file on it. Several dropped together are one note, because they arrived in one gesture.
• Onto a note — the file is attached to that note. The note lights up as you hold the file over it and says so, so you can see which one it is going to before you let go.

Paste a picture and it stands in the note where you pasted it, and is attached as a file as well. Copy an image or take a screenshot, then paste it while you are writing a new note, editing a note or with the edit panel open. The picture goes in at the cursor as its own block between paragraphs: words before it, the picture, then words after it. Pasted in the middle of a paragraph, it splits the paragraph there; there is always a line after it to go on typing. Its file is listed with the note's other files at the bottom, as it always was. Pasted on empty space, in any view, it makes a new note with the picture in it. Several pictures stand one after another, each its own file, named by the moment you pasted them unless they already had a name. A file copied beside a picture comes with it, and a file that is refused is refused with the same words as a dropped one, and nothing is put in the note. If what you copied also holds words or a table, as it does from Word or Excel, the words and the table are pasted and the picture is left behind.

• Show in note puts a picture in the words or takes it out. In the open note, each picture's row among the files has a Show in note switch. Turn it off and the picture leaves the words, which close up; the file stays. Turn it on and the picture goes back where it was, or at the end of the words if the words around its place have changed since.
• Dropped, attached and emailed pictures start as files only. A picture you drop on a note, add with Attach a file, email in or send from your own script is a file, as before, with Show in note off. Turn it on and the picture appears at the end of the words. To move a picture within its note, cut it and paste it where you want it.
• Taking a picture out of the words never deletes its file. Select it and press Backspace or Delete, or cut it, and it leaves the words only: the app says its file is still attached, and Undo puts it back where it was. Deleting the file, with × on its row, takes the picture out of the words too, and Undo brings back both.
• Copied to another note, only the words go. A note never shows a picture from a file it does not hold. To show the same picture in another note, paste or attach the picture there.
• Where you see it. Wall and Board cards show each picture at its place, as the note was typed; a long note still folds. List rows, the Calendar and folded cards show the title, and a title is made from words only. A picture whose file has not reached this device yet keeps its place with a box that names it, until it arrives.

A dragged link is still a link either way: it becomes a note with its preview, never an attachment.

• The file lives in this browser, beside the notes. Not on a server — the same place everything else is kept. It is copied to your own drive as well, so your other devices get it.
• A picture opens where you are. Tapping it shows the image over the wall, with a Save button if you want a copy. Nothing opens a new tab, which is a thing phones refuse to do anyway.
• A file that is refused keeps its row, struck through and marked, with the reason when you hover. Seeing WHICH of five files was turned away is the point; one quietly not appearing is the version that wastes an afternoon.
• Some kinds are never accepted — anything that could run. There is a size limit too, and both are settings rather than opinions written into the code.

Previews of attached files Works now

A note that is only a file looks like the file.

When a note has an attachment and no words, no checklist and no link preview, its card shows a preview of the first attachment instead of a blank card. Add words and the preview goes: a note with words never shows one. A note whose only words are the name of its file, which is how a dropped file starts, counts as having none.

• Pictures. JPEG, PNG, GIF (its first frame), WebP, BMP and ICO show on every browser. AVIF, HEIC, TIFF and JPEG XL show where your browser can draw them, which for HEIC, TIFF and JPEG XL means Safari; elsewhere the card shows the file's type instead. What your browser can draw is tried, not assumed, and the file's own bytes decide its type, not its name.
• PDF. The first page. A PDF that needs a password, one that takes more than a few seconds to draw, or a browser that cannot draw it, shows the file's type instead.
• Word, Excel and PowerPoint. The small picture the file carries inside itself, when it has one. Most PowerPoint files and every LibreOffice file do; many Word and Excel files do not, and an older .doc, .xls or .ppt never does. That is how the file was saved, not a fault, and the card shows the file's type instead: its extension in large letters, its name and its size. No document is opened or run to make a preview.
• SVG shows the file's type card rather than the picture.
• More than one file. A +N badge counts the others. A folded card shows its title line only, which for a note with no words is the first file's name.
• Tapping it opens the file exactly as the attachment row does. A swipe that starts on the preview is still a swipe, and holding it edits the note, as on any card (see Opening a note).
• Large or missing files. A picture over 20 MB, or of more than 50 million pixels, shows its type card and is not decoded. A file your device has not yet received from your drive shows its type card until it arrives. In the List a row shows the file's name, as before.
• Made on your device, and kept there. The preview is drawn by your browser from the file on your device. It is not sent to our server, not synced, and not in a backup or an exported file. Deleting the file or the note, emptying the trash, or erasing the device removes its preview with it.

Mind maps Works now

Ideas as a map of topics, kept with the note and opened full screen.

Open a note's ⋮ menu and choose Add a mind map…, press Add a mind map under the files in the open note, or write /object mindmap. The map opens full screen with one topic, Central Topic, for you to rename. On the note it is listed with the files as Mind map · and the name of its central topic, with the number of topics; press it to open the map again. Close the map with × or your browser's Back button. What you change is saved as you go, and a map you close before changing anything leaves the note as it was.

• Adding topics. + Add topic adds a sub-topic under the selected topic, or under the central topic when none is selected; the + beside a topic does the same for that topic. Click a topic to select it, and double-click it to rename it (tap and double-tap on a touch screen). A selected topic opens the Topic panel, with its Label, Notes and Color. The panel's × closes it and leaves no topic selected, as Escape does.
• Moving topics. Drag a topic to place it by hand. Drop it onto another topic, which shows a ring, and it moves under that one; a topic cannot go under one of its own sub-topics, and a move like that springs back. In the panel, Move up and Move down change its place among its siblings, Move under… lists where it can go, the four arrows under Position move it by hand, and Put back in its place returns it to where the layout puts it. Beside Position the panel says whether the topic is Placed by the layout or Placed by hand.
• Layout. Layout ▾ arranges the map Both sides of the central topic, to the Right, or Top down. A topic you placed by hand keeps that place in each layout.
• Hiding a branch. The fold toggle on a topic with sub-topics hides them and shows them again.
• Notes on a topic. Write them in the panel's Notes; Markdown is supported. A topic with notes carries a mark, and holding the pointer over it shows them (a long press on a touch screen).
• Colours. Default, Green, Orange, Purple, Blue and Red, under Color in the panel; the chosen one carries a ✓. With the keyboard, Tab to the colours and the arrow keys change the topic's colour. The topics stay white with dark text in the dark theme too, so a map looks the same on paper.
• Undo and Redo. ↶ and ↷ in the toolbar, or Ctrl+Z and Ctrl+Shift+Z (Cmd on a Mac), up to 100 steps, while the map is open.
• Deleting. Delete topic removes the topic and everything under it, and Undo brings it all back. The central topic cannot be deleted. To delete the whole map, choose Delete this mind map under ⋯, or press × on its row in the open note; the message that follows has Undo.
• Zoom. − and + beside the percentage, and Fit map to view (⌖) to see all of it. On a touch screen, pinch.
• With the keyboard. The arrows move along the lines as they are drawn, and Home goes to the central topic. Enter adds a topic beside this one, Ctrl+Enter (Cmd+Enter) a sub-topic, F2 renames, Delete deletes, Space hides or shows sub-topics, Alt+↑ or ↓ (Option on a Mac) moves a topic among its siblings, and Shift+arrows places it by hand. Escape steps back one thing at a time, and then closes the map. Tab only moves between the toolbar, the map and the panel; it never adds a topic. ?, or Keyboard shortcuts under ⋯, lists every key.
• As an outline. Outline in the toolbar shows the map as an indented list, which a screen reader reads as a tree; press it again for the map. Layout, zoom and Fit map to view stay in the toolbar but work in the map only, and the panel has no Position there. Enter, Ctrl+Enter, F2 and Delete work there as on the map, and it stays an outline while you add and rename. The arrows go up and down the list, → opens a branch or goes into it, ← closes it or goes back to the topic above, End goes to the last topic, and typing a letter goes to the next topic that starts with it.
• Export. Export PDF opens your browser's print dialog with the map alone, on landscape A4; choose to save it as a PDF there. Export outline downloads the map as a Markdown outline, a .md file named after the central topic. Both are made on this device; nothing is sent.
• On paper. Print a view draws each map in Full notes, with its topics' notes under the drawing, and prints its outline in Compact.
• Search finds its words. The view's search looks in the topics of every map, and finds the note that holds it. The card itself does not draw the map.
• Part of its note. A map is sealed, synced and backed up with its note, goes to the Trash and comes back with it, and our server cannot read it. A map in a note in the Trash opens to read only; put the note back to change it.
• Changed on another device while open. If you have not changed it here, the map reloads in place. If you have, the copy saved later wins, normally the one you are editing, and the map says This map was also changed on another device. The version here is the one kept.
• How big. Up to 2,000 topics in a map, 200 characters in a label, 5,000 in a topic's notes, and 10 mind maps, process maps and data models together in one note. The map says so when you reach a limit, and nothing you already have is cut.
• On a phone. The Topic panel comes up from the bottom, short at first; More shows every field. Some toolbar buttons move under ⋯.
• A map from a newer app. If a newer version of the app made a map this one cannot read, it is kept untouched and listed as made by a newer version of the app, and it does not open until this device has the newer app.

Data models Works now

A data model written as plain lines, drawn for you, kept with the note.

Open a note's ⋮ menu and choose Add a data model…, press Add a data model under the files in the open note, or write /object datamodel. The model opens full screen with two panes: the Markup, where you write the model, and the Diagram, which draws it as you type. On the note it is listed with the files as Data model · and its title, with the number of entities; press it to open the model again. Close it with × or your browser's Back button. What you change is saved as you go.

A model is made of segments (areas of the business, such as Sales), the entities in them (the things the business keeps data about, such as Customer and Order), the relationships between entities, and each entity's attributes. You write one thing per line, and indent a line to put it under the one above.

title Shop
columns 2

segment Sales color purple
  Customer "a person who buys from us"
    * customer id: integer
    email: varchar(200) not null unique
  Order "a request to buy"
    * order id: integer
    placed on: date "when it was placed"
    table orders
    rule An order has at least one item
  Order Item

segment Catalogue
  Product
    * sku: varchar(20)
      column product_sku

Customer 1 places 0..* Order
Order 1 contains 1..* Order Item
Order Item references Product

Two segments side by side: Sales holds
Customer, Order and Order Item; Catalogue
holds Product. Lines join the entities,
each with its phrase: places, contains,
references.
No warnings

• The lines you can write. At the left edge: title and subtitle with their words; columns with a number from 1 to 12, the segments in a row (3 if you leave it out); segment with a name, and after the name, in any order, at row 2 column 1, span 2 (two cells wide) and color with one of blue, purple, green, orange, red or neutral. Indented under a segment, an entity's name, with its definition in double quotes if you like, and the segment's database, schema, partition by and replica. Under an entity, its attributes: a name, then : and a type, then a definition in quotes; a * in front marks the attribute as part of the entity's identifier. Also under an entity, table with its name in the database, and rule with a business rule. Under an attribute, column with its name in the database. A line that starts with # is a comment for you: it is kept and never drawn.
• Relationships. At the left edge, write two entities with the words that join them between: Customer places Order. Add how many next to either name, as 1, 0..1, 1..* or 0..* (or in words: one, zero or one, one or more, many). A relationship may come before its entities in the text. A new name at the end, such as Wallet in Customer owns Wallet, makes a new entity, put in No segment until you write it under one. For a kind of another entity, write Premium Customer is a Customer.
• Forgiving. Names match whatever their capitals and spacing, so order item is Order Item; the spelling drawn is the first one in the text. Indent with spaces or tabs, by any amount, as long as you are consistent. A name that would read as one of the words above, such as Title Deed, goes in double quotes: "Title Deed". An entity is safest under a segment: a line at the left edge that starts and ends with names you have used elsewhere reads as a relationship.
• Warnings. A line the app cannot use is marked beside its number, and the list under the markup says what is wrong in a sentence, with the line number. The rest of the model still draws, so a half-typed line never hides your drawing. The list's heading counts them, as No warnings, 1 warning and so on.
• Levels and presentations. Level chooses how much detail; Presentation chooses how it is shown. Conceptual has the Domain map (the drawing of segments, entities and relationships) and the Entity–domain matrix. Logical has the Data dictionary. Physical has the Database schema diagram and SQL DDL, the statements to create the tables. The model remembers the one you last chose. A table or text presentation has Copy and Download: a table downloads as a CSV file, the SQL as a .sql file. The app writes the SQL; it never runs it.
• Gaps. At the Logical and Physical levels the list also counts what the model does not say yet: entities without a definition and relationships without a cardinality, and at Physical also entities without an identifier and attributes without a type. Show in diagram on a gap selects the items it counts.
• Working in the diagram. Click an entity, a relationship or a segment to select it, and Details opens its panel; or hold the pointer on it for 3 seconds, or for 1 second to rename it. An entity's panel has its Name, its segment, its Definition, its attributes and rules, its relationships, and Relate to…. A relationship's panel has its Phrase, its cardinalities and Reverse. Every change you make in the diagram is written into the markup, on the lines it concerns and nowhere else. A change that would make the markup read back as something different is refused, and nothing changes. Show in markup goes to the item's line, and Alt+Enter in the markup goes from a line to its item in the diagram.
• Auto arrange. A new model has Auto arrange on: the app places every segment and entity, and dragging an entity moves it to another segment. Turn it off to place things by hand: drag an entity, or press Shift+arrows; Ctrl+Shift+arrows (Cmd on a Mac) resizes a segment. Places you give by hand are written at the end of the markup, after a comment line and a layout line, and stay there when you turn Auto arrange back on. Arrange now removes the hand places of the presentation on screen. To go back to the automatic drawing everywhere, delete the markup from that comment line to the end.
• Undo and Redo. ↶ and ↷, or Ctrl+Z and Ctrl+Y (Cmd on a Mac), up to 100 steps, while the model is open. One Undo covers both panes: it takes back typing and changes made in the diagram alike.
• Deleting. Select an item and press Delete, or use Delete entity, Delete relationship or Delete segment in its panel; the message that follows has Undo. Deleting a segment keeps its entities, in No segment. To delete the whole model, choose Delete this data model under ⋯.
• With the keyboard. In the markup, Enter starts a line with the same indentation, Ctrl+] and Ctrl+[ (Cmd on a Mac) indent and outdent the lines selected, and Tab leaves the field. In the diagram, the arrows, Home and End move between items, F2 renames, Ctrl+Enter adds an entity, Space folds a segment and Shift+F10 opens the item's menu. Keyboard shortcuts under ⋯ lists every key. A screen reader reads the diagram as a tree: segments, then their entities.
• Export. Under Export ▾: Export PDF (this presentation) opens your browser's print dialog with the presentation on screen; Export markup (.txt) downloads the markup exactly as written, which is the whole model; and at the Physical level, Export SQL (.sql). All are made on this device; nothing is sent.
• On paper. Print a view draws the presentation last shown in Full notes, with the markup after it, and prints the model as an outline in Compact.
• Search finds its words. The view's search looks in the title, the names, the phrases, the definitions and the rules of every model, and finds the note that holds it. Comments are not searched.
• Part of its note. A model is sealed, synced and backed up with its note, goes to the Trash and comes back with it, and our server cannot read it. A model in a note in the Trash opens to read only; put the note back to change it. If it was also changed on another device while open here, the copy saved later wins.
• How big. Up to 200,000 characters of markup, 1,000 entities, 60 segments and 3,000 relationships in a model. Past the entities, segments or relationships, the first line over is warned; it and any more of its kind are left out of the drawing, and the markup keeps every word you wrote. A definition or rule longer than 2,000 characters is shown cut, and kept whole in the markup.
• On a phone. Markup and Diagram are two tabs, and Indent and Outdent buttons sit under the markup. With Auto arrange on, the segments take fewer columns to fit the screen.

Twelve words Works now

Your recovery phrase, and the only way back in.

It is always on: Settings → Encryption and your devices shows it. Your notes are sealed before they reach our server, so it cannot read them, and your backups are sealed too unless you choose readable ones; the few exceptions are listed in the Privacy Policy. On the device itself, notes are kept the way your browser keeps any site's data, so protect the device with a passcode and, on a laptop, disk encryption.

• You get twelve words. They are the only way back in. We do not have a copy and cannot make one — if you lose both the device and the words, the notes you already have are gone. You can still start again for new ones: see I lost a device, or my twelve words.
• Changing the key needs the key. Signing in is not enough. Whenever a device replaces your key, or gives it new words, the server first asks it to open something only your key can open — so somebody who got into your Google or Microsoft account cannot point your incoming notes at a key of their own. You will not see this happen; a device that holds the key just answers.
• They cannot be shown again. They are not kept anywhere — not in this browser and not on the server — which is exactly what makes them worth writing down. Settings will tell you this rather than pretending to have them.
• Lost them? Nothing is gone, as long as one device still opens your notes. The phrase does not hold the notes; it wraps a copy of the key so that ANOTHER device can be given it. A device that is already set up has the key itself. Go to Settings → Recovery phrase → Set a new phrase and the same key is given new words: every note stays readable, your backups still work, and your other devices carry on untouched. The old words stop working, so destroy any copy of them — and a device you have not set up yet will need the new phrase.
• Dates, the category and a few other details stay readable while a note waits on our server. Its words, #tags and sub-tags are sealed together. Your devices need the rest to sort and count what is waiting before opening it, and the default tags you choose for each way in stay readable too. A sealed backup seals all of it. Put the detail in the note, not in the category.
• On the web the key lives in browser storage, which is less protected than a phone's keychain. It leaves your browser only locked by your twelve words: inside your backups and exported files, and as the copy our server keeps so that another device can be set up with the words.

Backup and restore Works now

A copy of everything, in your own cloud or on this device.

Your notes live on your devices. Our server holds a note only until your devices collect it, and never more than 30 days, which is good for privacy and no help at all if the phone goes in a river. A backup writes the whole library into your own cloud as a single file: Google Drive if you signed in with Google, or OneDrive if you signed in with Microsoft, in a folder of its own that only this app uses. Or onto this device, if you would rather not connect a cloud at all.

MyNotesHelper/
  mynoteshelper-<account>-backup.json
  mynoteshelper-<account>-backup.1.json
  mynoteshelper-<account>-backup.2.json
  mynoteshelper-<account>-file-<id>.bin

the one just written
the one before it
the one before that
one attachment, one file

• It is sealed unless you choose otherwise. Only your devices, or someone with your twelve words, can open it. If you would rather have backups any program can read, choose that in the backup drawer: the app asks first and records the choice, and new backups are readable from then on. Export a readable copy writes one readable file whenever you want, whatever that setting says.
• A saved file holds everything, attachments included. Export to a file writes one complete file — every note, every attachment and your key, locked by your twelve words — so there is nothing to collect separately and nothing that can go missing on its own. That is what makes it the copy to keep somewhere safe.
• In your Drive or OneDrive, the attachments sit beside it. Each one is its own small file in the same folder, written once and never rewritten. The backup names them; it does not carry them. Each is sealed the way the backup is, and readable only if you chose readable backups; a file this device cannot seal is not sent.
  
  This is why syncing is quick. When attachments travelled inside the backup, changing one word in one note re-uploaded every photograph you had ever attached, three times over — once for each copy kept. Now a photograph is uploaded the first time and never again, and a note edited today sends the notes and nothing else.
  
  Nothing is lost when a note goes: an attachment stays until no kept backup mentions it any more, which is the same promise as before.
• It is read back before it is called done. Every backup is fetched again and counted. A write that silently wrote nothing is the failure nobody notices for six months.
• The new copy is written and checked BEFORE the old ones move. A connection lost halfway leaves every copy you already had exactly where it was.
• The newest is always the same filename. So there is one obvious file to restore from. See Copies to keep for the older generations.
• Restoring never deletes. A note you wrote since the backup survives it. Where both copies exist the newer one wins, so restoring twice does nothing the second time.

Restore takes it from your Drive when one is connected — the app knows which file it wrote — and otherwise asks you for a downloaded one. That second route is there for moving between accounts, coming back from an export, or restoring on a machine that has never been connected.

Download my data Works now

Everything our server keeps about your account, as one file.

Open Settings → Your data and press Download my data. You get a file named mynoteshelper-my-data-the date.json, in a form other programs can read. It holds your account details, settings, plan and payments, linked numbers and addresses, and reminders, as our server keeps them. Notes still waiting for your devices are in it locked with your key. Your full library is in your backup.

• You sign in again first. The file holds your email address and phone numbers, so a session that is already open is not enough. If you signed in within the last ten minutes you go straight to the download; otherwise Confirm it’s you asks you to press Sign in with Google or Sign in with Microsoft, whichever your account uses, and the download starts when you come back.
• Only you can collect it. There is nothing for us to send, and we never send account data by email. If you cannot sign in, write to mail@logoslegacy.com and we will tell you how to get back into the account.
• It says what it leaves out. Keys and security records are not in the file: the key locked with your twelve words, key-change checks, session identifiers, API key secrets, push addresses, rate counters, retry records, and diagnostics and channel logs. The file lists what is left out, and you can ask us about any of it.
• Once a day. A second download the same day says so and asks you to try tomorrow.

Syncing by itself Works now

A note written on one device, on the other one a minute later.

Some accounts sync through the MyNotesHelper server rather than through a Drive of their own. Those accounts do not have to press anything: the app checks for changes on a timer, brings down whatever another device wrote, and puts it on the page where you are.

If you back up to your own Google Drive or OneDrive, this does not apply to you and Sync now is still how it happens. That is deliberate rather than an omission — see below.

• Nothing changes under your hands. The page is never reloaded and nothing redraws while you are typing, dragging a card, or have a panel or a menu open. Notes that arrived in the meantime are already saved on this device; they appear as soon as you are not in the middle of something. Where you had scrolled to is kept.
• A sync that finds nothing changes nothing on the page. No panel and no banner. The account button says Syncing… while it runs, then when this device last synced. The point of it is that you stop thinking about syncing at all.
• It works in both directions, and deletes nothing. It is the same merge Sync now has always run: where two devices both have a note, the newer one wins.
• Deleting is the last word. A deleted note goes from your other devices rather than reappearing, and it stays gone even if one of them had touched the note more recently — opening a page it linked to, or adding a tag, counts as touching it. Nothing brings a deleted note back except you taking it out of the trash, and doing that puts it back everywhere.
• It waits until you are looking. A tab in the background does not sync, and syncs the moment you come back to it — so a phone in your pocket is current by the time you have unlocked it, without having spent the night polling.
• It needs your twelve words on this device. Everything that reaches the server is sealed, and a device that cannot seal cannot sync. Enter them once and it starts on its own.
• It tries again later. A failure turns the dot on the account button red, beside the time of the last sync that worked, or beside Not syncing if this device has never synced. Your notes are on your device and nothing is at risk, and it waits longer after each one rather than hammering a server that is already having a bad minute. The details are in Diagnostics.
• Why only these accounts. Your Google Drive is yours. Reaching into it every minute unasked is not the arrangement you agreed to when you connected it, it can raise a permission window you did not press anything to get, and it spends a quota that belongs to you. Our own server has none of those problems, so the automatic sync is offered exactly where it costs you nothing.

How often is set by whoever runs this deployment and is the same for everybody on it. There is nothing to configure and nothing to switch on.

When a backup fails Works now

What each failure means, and what the app does about it.

The backup panel reports what actually happened to your last backup, in the words of whatever refused. This is the reference for what those words mean.

• Your Drive is out of space. Nothing is changed — the previous backup is still there and still whole. Download the file instead and free some space when you can.
• Access to your Drive has gone. Revoked, or the folder was deleted. Your notes on this device are untouched, and reconnecting is enough to put it right.
• Google blocked the window. The first backup asks Google for permission to write one file, which is a separate approval from signing in. If the window is refused, allow pop-ups for this site and press the button again.
• The file did not read back as a backup. The half-written file is removed and your previous backups are left untouched. Download the file instead, and keep it.
• Drive failed for any other reason. Nothing is saved by itself. The panel says the backup failed rather than showing a tick, and offers Save a file instead, so you can keep a copy on this device while you sort it out.

What the server can read Works now

Stated plainly, including the exceptions.

Your notes are encrypted before they leave your device, so the text of a note and its #tags and sub-tags are not something this service can read. The details around a note are the exception: its dates, its category, an attachment's file type and size, and the default tags you choose for each way in are left readable, so your devices can sort what is waiting without opening everything first. Bear that in mind when naming a category: put the detail in the note.

A few things you choose to do send a note's words and tags out: /summarize, /ocr and a question to the assistant. All three are switched off until you switch them on and accept a short explanation — see Switching the AI features on and off. Each says so where you use it. Apart from those, notes reach our server sealed, with the exceptions listed in the Privacy Policy, such as the dates and category, the links our server fetches to show a preview, and the titles of reminders you choose to receive by text or email. Reminder phrases are read on your device. Nothing is sent.

The Privacy Policy says exactly what is and is not readable, and by whom.

A note I sent has not arrived

By WhatsApp, text message or email.

• Is that way in connected? Open Settings → Capturing from elsewhere. The button reads WhatsApp connected, SMS connected or Email connected once it is. A message from a number or address that is not connected is not saved, and nothing is sent back to say so.
• Is it connected for the way you sent it? A phone is connected for text, for WhatsApp, or for both — the list of your phones says which, beside each number. A WhatsApp message from a phone connected only by text is not saved, and neither is a text from one connected only by WhatsApp. Connect the same phone again with a code for the other way in and it is added, so both then work.
• Has this device collected it? Open the account menu at the top right and choose Check for new notes. Nothing waiting. means the server has nothing for you. If it asks for your twelve words, enter them first: anything waiting is sealed to them.
• Email takes longer. The server collects mail on a schedule, not the moment it lands: four times an hour, at 2, 17, 32 and 47 minutes past. Give it a quarter of an hour before you worry.
• An email must be shown to be yours. It is saved only when your mail provider has signed it for the connected address. If mail from your address never arrives, your domain may not sign its mail: connect an address at a large mail provider instead.
• Text and WhatsApp spend credits. When they run out, notes sent from your phone do not arrive and are not queued. Settings → Your plan says so when it happens.
• It will not wait for ever. A note waits on the server for up to 30 days for one of your devices to collect it.
• An administrator may clear it sooner. If the service's administrator removes items that were waiting for you, the app tells you, once, the next time it opens. The administrator sees how many items there were, how big and how old, and never what they said.

A reminder did not arrive

The notification did not come, or came late.

• Look at the bell first. It appears at the top when a reminder has come due, and it counts them whether or not a notification reached you.
• Is Push on? Open Settings → Reminders and look at Reminders arrive by. If your browser is blocking notifications, the app says so and turns Push off. Allow notifications for this site in your browser's settings, then turn Push on again.
• On an iPhone or iPad, a notification can only arrive from the app added to your Home Screen. See Reminders that reach a closed app above.

It says Not syncing

The account button shows a red dot beside the time it last synced, or beside Not syncing if this device has never synced.

• Your notes are still on this device. Nothing is lost while sync is failing, and the app keeps trying.
• Try it yourself. Open the account menu and choose Sync now. If a Google or Microsoft window asks for permission, give it. If the window never opens, allow pop-ups for this site.
• If it still fails, the Sync sheet says why, and offers Save a file instead. Save one, so there is a copy while you sort it out. When a backup fails explains each reason.
• Diagnostics has the details. Open the account menu and choose Diagnostics. What this device tried lists each attempt, and Copy all copies everything, for anyone helping you. Getting help says where to send it.

Setting up another phone or computer

Sign in, then prove it is you with your twelve words.

• Sign in the same way as before. A Google account and a Microsoft account are two separate accounts here, even with the same email address. Signing in with the other one opens an empty account.
• Enter your recovery phrase. Type the twelve words in order and choose Unlock this device. Capitals, numbering and commas do not matter.
• If the words are refused, the message says how many words you typed when there are not twelve, which word is not one a phrase can use, or that they are all real words but do not belong together (usually one was read or typed wrong), or that they belong together but are not this account's. That happens with an old phrase you have since replaced.
• Another device can walk you through it. On a device that is already set up, open Settings → Encryption and your devices and choose Add a device….

I lost a device, or my twelve words

What still gets you back in, in order.

• You have the words, and lost the device. Sign in on another one and enter them, as above.
• You lost the words, and a device still works. On that device, open Settings → Encryption and your devices, choose Show it beside Recovery phrase, then Set a new phrase. Write the new words down. The old ones stop working.
• No words and no working device. A backup cannot set up a new device on its own: the key inside it is locked by your twelve words. The exception is a file you downloaded with Export to a file before backups were sealed, which carries your key unlocked. Sign in, choose I don’t have my words on the recovery screen, then Restore an older backup file, and pick that file. A newer backup is refused before anything changes, so trying costs nothing. Keep such a file somewhere only you can reach.
• None of those. The notes you already have cannot be unlocked by anyone, and that does not change. What you can do is start again for the new ones: on the recovery screen choose I don’t have my words, then Start a reset. You are given twelve new words — write them down — and after a waiting period the app shows you, notes arriving from then on open with the new key. Everything from before stays locked.
• A reset is slow on purpose. Until it takes effect, everything arriving is still sealed to your old key, and every device on your account shows a red banner with the date and a Cancel it button. That is the protection: anybody who took over your Google or Microsoft sign-in could ask for a reset too, and the wait is what gives you the chance to stop them. Cancelling needs a device that holds your key, or your twelve words — and it signs every other device out.
• If you see that banner and it was not you, cancel it, then secure the account you sign in with. Somebody who can still sign in can ask again.
• Notes this device cannot open. Notes locked with a key this device no longer has, usually from before your twelve words were changed, are listed in Settings → Encryption and your devices under Notes waiting for a key, with See them. That row is there only while there are some. They open on their own if this device gets that key back. Nothing else can open them.
• Keep the words on paper, somewhere you would look.

Sign-in does not work

What each message means.

• Sign-in was cancelled. The sign-in window was closed before it finished. Start again.
• Google sign-in could not load. Check your connection. A browser extension that blocks scripts or pop-ups can also stop it.
• That sign-in attempt expired or was already used. Go back to the sign-in page and start again. An old tab or a reused link causes it.
• Could not reach the server. The service did not answer. Try again in a few minutes.

Which browsers it is tested in

Google Chrome on a computer. Others are best effort.

• Tested: Google Chrome on a computer, including in a phone-sized window. The app's automated checks run in it. Microsoft Edge is built on the same engine and should behave the same, but the checks do not run in it.
• Not tested yet: real phones and tablets, Safari (on a Mac, iPhone or iPad) and Firefox. Most things should work, but nothing checks them. If something looks wrong there, try Chrome on a computer, and tell us what you saw.

Getting help

Write to mail@logoslegacy.com.

A person reads it. This is a small service with no support team, so a reply can take a while.

• Say what you were doing and what you expected to happen, and copy any message the app showed, word for word.
• Say which device and browser you were using.
• Paste in Diagnostics. Open the account menu, choose Diagnostics, then Copy all. It includes the build you are running and what this device tried. It records no message text.
• Never send your twelve words, a password, or the text of a private note. Nobody needs them to help you.

Free

Not a trial. It does not run out.

Your notes are never limited. Write as many as you like, keep them for as long as you like. They are never deleted, never expire, and are never held back from you.

Free includes writing and searching, every view, the wall, tags and sub-tags, sealing before your notes leave your device, backup and restore to your own cloud, sending notes in through the API or by email, and sending notes in from your phone by WhatsApp or by text.

WhatsApp is the cheaper of the two, and by a long way on a long note. A message from your phone costs credits because it costs us real money to receive — a network charges us for every text, by the 160 characters, and charges more again for a picture. WhatsApp is one flat price however long the message is, the same in every country. The plans page has the rates.

The assistant works on Free too. You start with a one-off handful of credits — enough for a good while of asking. It is not hidden behind a subscription: when the credits run out it says so, and everything else carries on working.

Two things are counted, and only two: tags and sub-tags. The plans page says how many.

In total across everything you have written — not per note. Putting a tag you already use onto another note costs nothing; only a brand new name counts.

Two things are not on Free at all:

• Saving your own views — the wall, every filter, every sort and every search work exactly as they do on Premium. What Premium buys is keeping an arrangement as a view of its own and switching between several of them.
• Reminders — being told about something later

WhatsApp used to be on this list and is not any more. It was held back as the expensive option, which had it backwards: it is the cheapest way a note can reach us. Keeping it behind a subscription only pushed people onto texting, which costs everybody more.

As with tags, views you already made are never taken away. If you built six on Premium and your subscription ends, all six stay — you can use them and delete them, and you simply cannot make a seventh.

Either way you have to connect a number first — that is what proves the phone is yours, and it works the same for both. After that, every message you send in spends credits, and WhatsApp spends fewer of them.

Credits

Credits pay for what costs real money: receiving a note by text or WhatsApp, where a photo sent by text costs more than words; the assistant, which means a question (one price, whatever it sends), a question about your whole library (priced by how much it sends, see Asking about a lot of notes), /summarize and /ocr; and, on Premium, a reminder sent to you by text or WhatsApp. Email costs nothing, in either direction. Everything else on the list above is free on every plan and always will be.

A text is charged by length and WhatsApp is not. Networks bill us per 160 characters, and a single accented letter drops that to 67 — so a long note in Portuguese can cost several times the same note in English. A WhatsApp message is one price whatever its length. That is the whole reason WhatsApp is cheaper here, and it is why we would rather you used it.

A free account starts with a one-off grant that never expires, Premium includes an allowance for each period you pay for, a month or a year, and more can be bought at any time.

The Premium trial includes a smaller allowance of its own, once per account. An allowance you have not used ends with its period, a few days after the end if the subscription is not renewed; credits you bought never expire. If a payment is refunded or charged back, what it bought is taken back: part of it for a partial refund, all of it for a full one.

What a credit buys, and what each plan costs →

Premium

A monthly or yearly subscription, with a free trial.

Unlimited tags and sub-tags, saved views, a monthly credit allowance for the assistant, and reminders. Everything on Free stays exactly as it was.

What it costs, and what is included →

Reaching a limit

Nothing is taken away. Ever.

When you are at the tag limit and write a note carrying one more, the note is kept and nothing is lost — only the new tag is left off. You are told which tag was not added and how many Free allows, and offered to reuse one you have, remove one you have finished with, or go Premium. A note sent in by the API is treated the same way: it is saved without the tag, and the answer says so (see The write API, in full).

If you were over the limit and your subscription ends, you keep every one of them. Every tag stays where it is. You can still search them, filter by them, and remove the ones you no longer want. The only thing you cannot do is add a new one until you are back under the limit or subscribed again.

This is deliberate. A plan that deleted your work when you stopped paying would be a plan that ate the thing you trusted it with.

23 October 2026, afternoon

Pictures where you paste them, previews that recover, and fixes.

• A pasted picture stands where you pasted it, with words before and after it, and its file is still listed at the bottom. Each picture's row has a Show in note switch. See Attach a file.
• Link previews recover. One that failed for a passing reason tries again by itself, and Try again does it now. Refresh link preview in the note's menu looks a preview up again. A preview is now kept until no note holds its link, and a device makes up to 100 automatic previews an hour for mail you send yourself. See Link previews.
• A note you have just written deletes or archives on the first press after a swipe.
• Mind maps: the + on a topic adds a sub-topic again, and a new topic lands beside its siblings instead of over another branch.
• List: a collapsed note no longer shows a stray line.
• Board: the column and card menus work from the keyboard.

23 October 2026

Mind maps, and files, views and the Board fixed.

• Mind maps. Choose Add a mind map… in a note's ⋮ menu, or write /object mindmap, and a map of topics opens full screen. It is kept with the note, has an outline view, exports to PDF or a Markdown outline, and search and printing reach its words. See Mind maps.
• × on a file in the open note removes it again, with a mouse or a finger.
• Every view can be reached in the view bar: arrows appear when the views do not fit, the mouse wheel scrolls the bar, and Tab brings a hidden view into sight. Settings set to open on the last view used now reopen a saved view after a reload too. See Views.
• Word, Excel and PowerPoint files with macros (.docm, .xlsm, .pptm) can be attached. Other macro and add-in files are still refused.
• Board: each column's buttons say which column they belong to when read aloud, focus stays put after you fold a column, and a double click on a note that is only a file edits it where it is.

21 October 2026

Board columns move by dragging their header.

• Drag a Board column by its header to put it somewhere else, the way you drag a view along the view bar. With a mouse, move sideways; with a finger, rest on the header for half a second, then move. Escape puts it back, and near the Board's edge it scrolls along with you. See Views.
• The new order belongs to that view and is the same on your other devices. Move left and Move right in the column's menu still do it from the keyboard.
• Dragging cards and panning the Board work as they did.

20 October 2026

Stripes on the right lines and easier to see, and a Refund Policy.

• /stripes shades the right lines. A line with words and the command on it counts; a line holding only a command does not. See /stripes.
• The stripe is easier to see, a warm tint in light and dark, and every word and tag on it stays readable.
• A Refund Policy page, linked at the bottom of every public page, says when you can have your money back. See the Refund Policy.
• Credit packs have their prices on the plans page, and the header has a Pricing link. See the plans page.

20 October 2026

Checklists that start where you write them, and striped notes.

• /checklist keeps your lines in the order you typed them. Before, the command's own line jumped to the top.
• /checklist lower down in a note starts the list on its own line. The lines above it stay ordinary text, and the list runs to the end of the note. See /checklist.
• /stripes gives every second line of a note a light band behind it, on the wall and board cards and in print. Your words, tags and links keep their colours. Delete the word to stop it. See /stripes.
• A command picked from the menu keeps its space, so you can type straight on after it.
• Lime tags are a shade darker in the light theme, so they stay readable on a stripe.

16 October 2026

Tags can be added on a Board again.

• On a Board, + tag on a card keeps its list open until you pick a tag or leave it. Before, the list closed at once, so no tag could be added there. The Wall and the List are unchanged. See #parent/child.

16 October 2026

New notes from every view, merging sub-tags, and swipe buttons that go away.

• Every way of making a note works in every view. The + button, the N, + and C keys and pasted words now make a note on a Board, the List and the Calendar too, starting with the view's tags. See The + button.
• A new note stays first while you are making it. Pinned notes no longer push it down while you type or add tags; it moves to its place once, when you move on.
• Merge tags works for sub-tags, under one main tag or across two, and between a sub-tag and a main tag. The words in your notes change with it, archived and trashed notes included, and Undo puts it all back. See Merge tags.
• An Archive or Delete button no longer stays on the screen after its card has gone back. See Archiving a note.
• Sorting by last edited follows your edits. Before, it gave the same order as newest first.
• Notes waiting on the server are never lost when a page opens without its encryption script; they wait and are opened once your key is there.

13 October 2026

A Board you can move around with a mouse, and new notes that stay in the view.

• A Board moves with a mouse, left, right, up and down. Drag on its empty background, use Shift and the wheel, or the arrow and Page keys once it has focus. Hold a dragged card at the Board's top or bottom edge and it scrolls. Dragging cards and touch work as before.
• A note you make in a view with tags picked starts with those tags, so it stays in the view. A message names them, with Undo. See Views.
• + Add card on a Board and + Add row in the List make a note again. Before this, they made nothing.

13 October 2026

No browser header on a printout.

• Chrome and Edge no longer print their own date, time and page title at the top of a printed view, whatever their Headers and footers option says. Our footer and the page numbers are unchanged. Firefox and Safari keep their own option, which you switch off in the print dialog. See Print a view.

12 October 2026

A tidier printout, pasted pictures kept as files, and Board order kept.

• A printout's footer has the date and then the time, and the top of the page no longer repeats them or the view's name. On a Board, each column's name starts with a circle in its colour. The print sheet remembers what you chose. See Print a view.
• Paste a picture and it is attached as a file, as if you had dropped it, while writing a new note or editing one in any view. Pasting from Word or Excel works as before. See Attach a file.
• On a Board, Unassigned keeps the order you give it, and offers Move up and Move down.

12 October 2026

Print a view, sub-tags with a main tag's name, and clearer link previews.

• Print a view. The Wall, the List, a Board or the Calendar, as Full notes or Compact, with formatting, checklists, links and small pictures. Press Print on the strip, use Print this view in the view's settings, or Ctrl+P (Cmd+P on a Mac). See Print a view.
• Printing your twelve words no longer leaves them in an open window afterwards. See Twelve words.
• A sub-tag can be renamed to a main tag's name. They stay two different tags, told apart by the full path. See Main, pinned, and everything else.
• A link whose page cannot be read now shows a card that says so, instead of nothing. See Link previews.

11 October 2026

Sync says what it is doing, everywhere at once.

• The account button, the Sync sheet and Sync now agree. Each says Syncing… while any sync runs, and the sheet shows the real step and how long it has taken each time you open it. See Syncing by itself.
• A press of Sync now shows at once, and the account button never stays stuck on Syncing… after a sync has ended.
• The same page open twice does not start a second sync.

10 October 2026

Sync keeps going when you close it.

• Sync now keeps running when you close the Sync sheet. Open it again, or press Sync now, and it shows the sync already under way instead of starting a second one. A sync that was interrupted carries on. See It says Not syncing.

10 October 2026

The List shows every note, and link previews say what they are doing.

• The List shows all your notes when no tag is picked. Before, on a large library, it could count the notes and show none of them until a tag was pressed. See Views.
• A note you email from your own verified address previews its first link on its own. Other emailed notes show Show preview, and a page that cannot be read says so on the card. See Link previews.
• Every card is spaced the same way, so stray blank lines no longer leave gaps.
• An address written without https:// becomes a link only when it ends in a real domain ending.

9 October 2026

An assistant you can say no to, formatted notes, file previews, and the gestures fixed.

• The AI assistant asks first, and goes away if you say no. A short explanation with equal Reject and Accept buttons. Reject and the assistant is not shown anywhere. Accept and it answers properly, with exact counts. See Switching the AI features on and off and It answers what it can itself.
• Notes can be formatted, and show it. Bold, lists, headings, tables and links in the open note, kept when you paste from a document or a web page, and drawn on the wall and board cards. See Formatting a note.
• A note that is only a file shows the file. A picture, the first page of a PDF, or a document’s own thumbnail, made on your device. See Previews of attached files. A note that is only a file can now be edited.
• Swipe works the first time, on the wall, the board and the list. See Archiving a note.
• Holding a card for three seconds no longer opens an empty, locked editor saying the words were changed somewhere else.
• Holding a card with a finger on the Board and the List opens the editor as it does on the wall: about a second edits in place, about three opens the side panel and keeps it open. See Opening a note.
• A deadline set in the Calendar no longer writes stray symbols into the note or drops its /due.
• A #tag you write is coloured the same in every place you can write one.
• Notes collected from email or the API show their words and tag, not the sealed wrapper they travelled in, and the ones already showing it were repaired.
• A note deleted on another device now also removes its files from this one.

8 October 2026

Rich text, a smarter assistant, your data in one file, and safer mail.

• Rich text in the open note, and a card dropped anywhere in a board column. See Views.
• Remove a tag from every note and delete it, from Main tags. See Main, pinned, and everything else.
• The assistant chooses the notes itself, can look up more while it answers, and tells you under each answer how many notes it was based on and what it cost. See Asking about a lot of notes.
• Download my data, in Settings. See Download my data.
• Save as a new view and Delete view. See Views.
• Sign out everywhere else, and a message on a device that was signed out from another one saying so.
• Email: strict checks on who sent it; too many in an hour are kept for you to bring in; links in an emailed note wait for Show preview. (Changed since: a note you email from your own verified address now previews its first link on its own; see Link previews.) See Email a note in.
• Backups are sealed by default, attachments are written beside them as their own files, and Restore an older backup file is on the recovery screen, under I don’t have my words. See Backup and restore.
• Delete these for good on ticked notes in the trash.
• Deleting your account also cancels a Premium subscription at once.
• Reminders and email are looked for every 15 minutes. See Reminders that reach a closed app.

30 September 2026

Safer ways in, and a cleaner account deletion.

• A phone is connected for the ways you connected it. WhatsApp and text are proved separately, and a message arrives once. See WhatsApp or text a note in.
• An email must be shown to be yours. Mail is saved only when your provider has signed it for the connected address, so a forged sender is turned away. See Email a note in.
• A note that cannot be sealed is not kept. Anything sent in that the server cannot seal to your key is refused rather than stored readable.
• Starting again without your twelve words now waits. A reset takes effect only after a waiting period, and every device shows a banner you can cancel it from. See I lost a device, or my twelve words.
• Deleting your account also removes your connected phones and email addresses and signs out every device.
• The Privacy Policy and Terms of Use are now version 3.1.