How Orchid thinks.

The mental model behind the product — what runs where and why.

The pages

Five short essays. Reference docs tell you what to click; these tell you why a thing is shaped the way it is.

  • How Orchid works — the three layers (IDE, kernel, cloud), how they talk, and why the seams matter.
  • Local-first — what stays on your machine, what you choose to publish, and how this differs from cloud notebooks like Hex, Deepnote, and Mode.
  • The .orchid format — why notebooks are YAML, how outputs spill to disk, and why this is the right shape for analytics work in git.
  • Projects on disk — what a project actually is, what lives in .orchid/, and how publishing flips a project from local-canonical to cloud-canonical.
  • Agent architecture — single vs multi-agent mode, the tool registry, and why every write pauses for review.

If you only read one

Start with How Orchid works. It frames everything else. Then read whichever page scratches the itch you came in with — they don't depend on each other.

Looking for how-to

For step-by-step walkthroughs, head to the guides and tutorials. These pages are the "why"; those pages are the "how."