Skip to main content
ZeroWidthdocs
Get started

Reference cards & connectors

Most whiteboard diagrams are drawings of things: a box labeled "triage flow" that drifts out of date the moment the real flow changes. In Napkin, you can put the thing itself on the board. A reference card is a live pointer to a workspace entity — it shows the entity's real name, opens the real entity when you double-click it, and reads as the real object when zv1 looks at the sketch.

What you can drop on a board

  • Flows (Workbench)
  • Pages (Compass)
  • Datasets and evals (Caliper)
  • Decisions (Ledger)
  • Other sketches (Napkin) — these render with a live thumbnail of the referenced board

Adding a card

Two ways, same result:

  1. Press @ and search

    From anywhere on the board, @ opens the entity picker. Type a few characters, arrow to the hit, press Enter — the card lands at the center of your view, selected and ready to drag into place.

  2. Paste a URL

    Copy the address of a flow, page, dataset, eval, decision, or sketch from its own tool and paste it onto the board. Napkin recognizes suite URLs and turns them into cards. (A link from a different workspace won't resolve here — you'll get a warning.)

Cards move, resize, and rotate like any shape, and the eraser removes them like any shape — removing a card never touches the entity it points to.

Cards stay honest

  • Names refresh. Each time the board opens, cards re-check their entities, so a renamed flow shows its current name.
  • Access is respected. A card only resolves for people who could open the entity themselves. If a teammate can't see the dataset behind a card — or it's been deleted — the card renders as missing rather than leaking anything about it.
  • Double-click to go. Sketch cards open the referenced sketch right in Napkin; everything else opens in its own tool.

Connectors

The connector tool (C) draws lines that bind to shapes. Drag from one shape toward another — a ring highlights what the endpoint will attach to — and release. From then on the connector follows: move either end's shape and the line re-routes itself. That's what keeps a diagram legible through the rearranging that real thinking involves.

  • Connectors bind to anything — sketched boxes, sticky notes, images, reference cards.
  • Either endpoint can also be free (a fixed point on the canvas) — bind one end to a card and leave the other end on a hand-drawn annotation.
  • Select a connector to toggle its routing between elbows (right-angle runs, the architecture-diagram register) and straight.

Why live beats static

A diagram made of reference cards is a working surface:

  • It's navigable. In a review, double-click straight from the sketch into the actual flow or eval being discussed.
  • It doesn't rot. Renames ripple in; deleted entities show up as missing instead of silently lying.
  • zv1 reads it with real names. Ask the assistant about the sketch and it sees which actual entities the arrows connect — not just anonymous boxes.
  • It travels. A Compass page can attach a sketch, so the diagram sits alongside the workflow documentation it explains.

Where to go next

3 min read