TakeSix

FAQ

Questions about TakeSix

TakeSix is an open format and libraries that let an AI agent answer with a diagram you can click into instead of a wall of text: at most six blocks per screen, and any block opens into up to six more.

What is TakeSix?

TakeSix is an open format and libraries that let an AI agent answer with a diagram you can click into instead of a wall of text: at most six blocks per screen, and any block opens into up to six more. The gist of the answer fits on the first screen; the reader clicks a block for depth, and long text opens next to the block that needs it.

How do I get TakeSix diagrams from my AI agent?

Connect the MCP server @takesix/mcp (for example, in Claude Code: claude mcp add --scope user takesix -- npx -y @takesix/mcp), or install the Agent Skill from the takesix-agent repository. Then ask for an overview, a plan or an architecture "as a TakeSix diagram". It works with Claude Code, Claude Desktop, Codex, Cursor, VS Code and other MCP clients.

How is TakeSix different from Mermaid?

Mermaid turns a text description into one flat picture, and a big subject makes a big, crowded picture. A TakeSix diagram is a tree of screens: each screen has at most six blocks, and any block opens into its own screen. The format is JSON made for agents to write: a validator lists every problem with its place in the file, so the agent fixes its own answer.

How is TakeSix different from the C4 model?

C4 is a way to describe software architecture at four fixed levels: context, containers, components and code. TakeSix is a general format: any subject, as many levels as it needs, and a hard limit of six blocks per screen. A C4 model fits into TakeSix well; the architecture blueprint of the agent skill follows the same idea of zooming in.

How is TakeSix different from draw.io or Excalidraw?

draw.io and Excalidraw are drawing tools: a person places shapes on a canvas by hand. In TakeSix an agent writes the diagram as JSON and places blocks on a 6×6 grid, so there are no coordinates to tune and every diagram reads the same way. A person can still open and change it in the TakeSix editor.

Why six blocks?

A reader takes in a handful of things at a glance. Six blocks are enough to say the whole thing on one screen and few enough to read without scrolling. When a subject needs more, it gets another level, not a bigger screen.

What does an agent produce?

A .takesix.json file and one self-contained HTML file that opens in any browser, without internet. In apps that support MCP Apps, such as Claude Desktop, the diagram also opens right in the chat.

Is it free, and where does my data go?

TakeSix is open source under the MIT license. The MCP server and the skill run on your computer; the diagrams are not sent anywhere. The editor on takesix.vapnik.de also works in the browser only.

Ready to try? Connect TakeSix to your agent or read the format.