Tutorial

Three levels of Tana commands

Work up from Tana's built-in commands, to building your own custom commands, to multi-step custom AI workflows including image generation and AI agents.

Almost everything you do in Tana is a command. This walkthrough builds up through three levels: the commands already built into Tana, the custom commands you can build yourself, and full custom AI workflows. If you'd rather skip the building, the Tana template store packages a lot of this up to add in one click.

Level one: built-in commands [0:00]

Every action in Tana is a command, surfaced in two places. The right-click context menu shows actions for whatever you're on, viewing in different formats, filtering, sorting, opening in a new panel, publishing, copying a link. The command line, opened with command + K (or control + K), reaches everything else: starting a meeting transcription, changing note formats, generating images (now upgraded), autofill, and the AI commands covered later. If there's one shortcut to learn, it's command + K.

Level two: building your own commands [9:33]

Type any text, then convert it to a command node. Inside, you stack system commands and set their parameters (insert a parameter with the > sign), for example a prompt, a target node, or a model. You can't place two of the same system command back-to-back, so nest them as separate custom commands to chain steps. Once created, a command is available everywhere in the command line, can be bound to a custom keyboard shortcut (shift + command + K), and can live as a button in a supertag's compact or full menu, like a high-priority button that sets status, priority, and due date in one click.

Level three: custom AI workflows [21:29]

The same building blocks unlock AI workflows. An audio-enabled tag with post-processing gives any node a microphone and autofills fields from what you say, the same mechanism behind the meeting notetaker. You can chain AI steps by feeding one command's output into the next through a field: for example, an ask-AI step writes an enhanced image prompt into a field, a generate-image step reads that field, and a final step removes it. At the far end are custom AI chats, an initial prompt, an agent with its own system prompt and source expressions, and command buttons in the chat. The template store ships working examples if you want to see how they're wired.

Three levels of Tana commands - Tana Outliner Learn