Notada has a local MCP server inside the app. Point the assistant you already use at it, and it can read your notes and write to them. The server talks only to your assistant, and nothing leaves your Mac.
Setup
The server is the notada executable inside the app, run with one argument, mcp. Where it goes depends on your assistant.
Claude Code takes it in one command, for every folder you work in:
claude mcp add --scope user notada -- /Applications/Notada.app/Contents/Helpers/notada mcp
Cursor takes it from a link: add Notada to Cursor, and confirm in Cursor. It installs the entry below, which you can also add to ~/.cursor/mcp.json yourself.
Claude Desktop keeps its servers in a file, which Claude › Settings › Developer › Edit Config shows you, and most other assistants have one of their own, in their own shape around the same two values. Add the entry below to that file. If the file already has "mcpServers", put the "notada" entry inside it, beside any others, and leave the rest of the file as it is.
Claude Desktop reads the file only when it starts, so quit it with ⌘Q and open it again.
Any assistant that runs commands on your Mac, Codex or Cursor’s agent as much as Claude Code, can do it for you. Paste this, and your assistant finds where it keeps its servers:
Set up Notada’s MCP server for me.
Notada is a notes app on this Mac. Its MCP server is the executable /Applications/Notada.app/Contents/Helpers/notada, run with the single argument mcp. It speaks over standard input and output, needs no environment variables, and makes no network calls. As a configuration entry:
"notada": { "command": "/Applications/Notada.app/Contents/Helpers/notada", "args": ["mcp"] }
First check that the executable exists at that path; if it does not, tell me and stop. Then register it under the name notada wherever you keep MCP servers, for me rather than for one project, since the notes are mine and not a project’s. If you have a command for adding a server, use it; if you keep servers in a configuration file, add this entry beside any others there and change nothing else in the file.
Then check that it connects, with your own command for that if you have one, and tell me whether I need to start a new session before its tools are there. macOS may ask once whether you may read my Documents folder, where the notes are; that prompt is for me to answer.
Then ask your assistant what’s open in my notes? It asks you before its first call to Notada, and macOS asks once whether it, or the terminal Claude Code runs in, may read your Documents folder, where the notes are. Say yes to both, and it is set. If you said no to macOS, turn it on under System Settings › Privacy & Security › Files and Folders, and open your assistant again.
The path is right once Notada is in your Applications folder. It works with assistants that run on your Mac; an assistant that lives only in a browser cannot reach a server on your Mac.
What it can do
Nine things, and your assistant learns them from the server itself:
List your notepads.
List the days you wrote on.
List your open todos, sorted as Notada sorts them.
List your open questions.
List your decisions.
List what you finished, dropped, or answered since a day.
Read the notes for a day or a range of days.
Append one line under today’s date, normalized the way the editor normalizes.
Close an open todo or question, which marks it done with today’s date.
The server never rewrites, reorders, or deletes a line. Your notes are a plain file, though, so when you ask your assistant to rewrite or tidy them, it edits the file itself, keeping your items and date headers as they are, and Notada reloads it. Your assistant is told that the list is derived from your notes, never kept by hand. If you keep more than one notepad, a question about what is open covers all of them, and each item says which notepad it is in; a line is only ever written to a notepad named, and your assistant asks you which.
Three commands come with it. /start_of_day starts your morning: what you finished on the last day you wrote, even when that was Friday, what is due today, the three oldest open items, and the longest-open question. /standup_brief writes a brief before standup: what you said you would do, what is due this week, and the longest-open question as a talking point. /carry_forward reviews the week: open items by age, decisions made, topics that came up three times with no decision, and promises in your notes that you never marked, offered as todos to add only if you say so. Your assistant lists all three as commands, each client in its own place; in Claude Code they are /mcp__notada__start_of_day and so on.
The one model Notada calls itself is the one built into macOS, on a Mac running macOS 26 with Apple Intelligence: it reads a line you wrote and may suggest that it is a todo, a question, or a decision, which you take or leave, and it never writes a line.
The command line
The same executable is a command line, for a terminal, a script, or an agent with a shell. Link it onto your path once:
Then it reads your notes, appends a line, or closes one. This is the front page’s notepad, as the command line reports it:
$ notada ledger --json
{
"decided" : [ … ],
"open" : [
{
"age" : "1w",
"ageDays" : 8,
"date" : "2026-09-03",
"due" : "2026-09-10",
"line" : 12,
"overdue" : true,
"state" : "open",
"text" : "Draft the definition of done with Alex"
},
…
],
"questions" : [ … ],
"today" : "2026-09-11"
}
$ notada append "- [ ] Reply to Anna about the trial length due:mon"
21: - [ ] Reply to Anna about the trial length due:2026-09-14
$ notada close 12
12: - [x] Draft the definition of done with Alex due:2026-09-10 done:2026-09-11
notada help says the rest. It reads the same file the app does, writes it the same way, and the window keeps up as it does with any change on disk. Questions go to hello@notada.app.
Notada