How BrainLog organizes writing, tasks, boards and sharing.
BrainLog has exactly three levels: Folders contain Sections, and Sections contain Pages. A page is a markdown document — everything you write is stored as plain text and remains exportable at any time.
Write a checkbox in any page and it shows up in that section's and folder's task list, board and due view:
- [ ] Send the quarterly report @due(2026-07-20) @assigned(alice) - [/] Draft the intro section - [x] Book the venue
- [ ] todo · - [/] doing · - [x]
done — the three board columns. Checking a box or dragging a card
rewrites that line in the page; there is no hidden task table it can
drift away from what you wrote.
Type + Board in the editor and three columns appear
inline. The columns are the headings, so renaming one renames
the heading and the two can never drift apart:
<!-- board --> ## Backlog - [ ] Rewrite the pricing page ## Done - [x] Book the venue <!-- /board -->
No delimiters, no board — however many headings a note has, an ordinary
note with ## Pros and ## Cons stays an
ordinary note. The delimiters are HTML comments, so the file still reads
as plain markdown everywhere else it's opened.
@due(2026-07-20) and @assigned(name) are part
of the task line itself. The Due view groups open tasks into overdue,
today, this week and later, and @assigned matches the
names of the folder's members.
.md files, mirroring its Folder/Section/Page structure.
If your notes are already markdown files, they can move in as they are:
open a section's ⋯ menu and choose
Import markdown to bring in .md or .txt
files — each becomes a page, titled from its first heading. Checkbox
lines in imported files are recognized as tasks immediately. And because
export is the same format, moving in is never a commitment: a zip of
plain markdown comes back out whenever you want it.
A folder can be created as end-to-end encrypted: check Encrypted when creating it, then choose a passphrase the first time you open one of its pages. Page content is encrypted and decrypted in your browser — the server stores only ciphertext, so the operator cannot read those notes. Page titles and the folder/section structure stay visible.
This is a deliberate trade. Inside an encrypted folder there are:
Normal folders keep every feature; use encryption only where the privacy is worth more than the tooling.
Each account can generate a token that lets AI tools — Claude Code, Claude Desktop, ChatGPT and other MCP clients — read and write the same notes: create pages, edit markdown, add tasks, move board cards. The connection cannot delete anything: the toolbox has no delete, trash, revoke or remove operation, by design.
Setup is one command. For Claude Code:
claude mcp add --transport http brainlog https://brainlog.net/mcp \ --header "Authorization: Bearer YOUR_TOKEN"
Generate the token — and find the guide for every other client, including the ones that only take a URL — under Settings → Connect an agent after signing in. Folders created with end-to-end encryption are invisible to every connected agent; what an assistant can reach is decided folder by folder.
Any folder can deliver its new pages to your inbox: open the folder's ⋯ menu (or Settings → Emails) and choose Each page or Daily digest. Every email carries a one-click unsubscribe.
Combined with a connected agent, this becomes a standing brief. Schedule a prompt in your assistant — Claude's scheduled tasks or ChatGPT's Tasks both work — along these lines:
Every morning at 7: summarize today's top headlines from the newspapers I follow, one line each with a link. Save it as a new page called "Headlines — <date>" in my brainlog folder "News".
With the News folder set to email you, the summary is in your inbox moments after the agent writes it — and the page itself stays in your vault: plain markdown, searchable, exportable, yours.