> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tatara.app/llms.txt
> Use this file to discover all available pages before exploring further.
# Document formats
> How a markdown document becomes a board, a tracker, a table, or a saved view, and what each format is for.
Every document in a brain is one of two file types: **markdown** or **HTML**. Almost everything you and your agents write is markdown.
A markdown document can optionally carry a **render format**. The format gives the document a richer view than plain prose: a board, a checklist, a grid, an index. The document itself stays markdown that anyone can read, diff, or paste into a chat window. Nothing is stored in a proprietary shape. The board *is* the markdown.
HTML documents are a separate file type, not a render format, and they never carry one. See [HTML documents](/formats/html).
## The four formats
There are exactly four render formats. They are fixed: a brain cannot define its own.
| Format | Pick this when |
| --------------------------- | ---------------------------------------------------------------------------------------- |
| [Kanban](/formats/kanban) | Work moves through stages and you want to drag it between them. |
| [Tracker](/formats/tracker) | You have a task list where due dates, priorities, and progress matter. |
| [Table](/formats/table) | The document is rows of typed columns, with optional computed values. |
| [Base](/formats/base) | You want a live view over the other documents in the brain, not rows stored in this one. |
A markdown document with no render format is an ordinary prose document and renders in the normal editor.
## How a document gets a format
Your agent assigns the format when it writes the document. There is no control in the web app that turns a document into a board: ask your agent for "a kanban board of the launch tasks" or "a table of our suppliers", and it creates the document in the right shape. A document keeps its format through later edits, moves, and renames.
## Why formats are strict
Tatara validates every write against the document's format, and the contract is a lossless round trip: the markdown source and the rich view must be able to produce each other exactly. If they could only almost produce each other, a person editing the board and an agent editing the markdown would slowly pull the document apart. So each format has one exact spelling, and an agent write that is close-but-not-exact is handed straight back corrected, ready to resend. What that means for you: the rendered view is never broken by a sloppy write, and the markdown underneath is always clean.
## What you see in the app
Open a document that carries a render format and you get the rich view instead of the text editor. Your edits (dragging a card, ticking a task, changing a cell) save automatically, the same way typing in a normal document does.
Three things are missing compared to an ordinary document:
* **No properties rail.** Owner, type, description, tags, links, and custom properties are not shown, and the shortcut that toggles the rail has nothing to toggle.
* **No title editing.** The title is displayed but read-only. Rename these documents from the sidebar instead.
* **No width control.** The Prose / Wide / Full setting is not offered.
If something is wrong with the document (an unknown format, an invalid config, or a body that will not parse), the app falls back to the plain markdown editor and says so in a banner:
```text theme={null}
This document is meant to render as kanban — showing plain markdown instead. (invalid config)
```
The parenthetical names the cause: an unknown format, an invalid config, a renderer error, or the first parse error from the body. You can always read and repair the document, whatever state it is in.
Columns and cards, dragged between stages.
Checklists with due dates, priorities, and tags.
Typed columns, with formulas.
Live views over the documents in the brain.
Static, self-contained HTML reports, created by an agent.