Canvas

Briefs

Real engineering boards, rebuilt as superlore Canvas — the kind of architecture, brainstorm, and review whiteboards a product team actually draws, authored as structured MDX.

These are not toy diagrams. They are the kind of boards a product-engineering team draws on a whiteboard while figuring something out — an agentic pipeline, a "today vs. where we want to be" review, a wall of asides and decisions. Each one is a single superlore-canvas spec: it renders as the board below and serializes to a typed node/edge graph an agent reads over the MCP.

Two things make boards like these possible beyond a tidy flowchart:

  • Not everything is a flow. A board can be disconnected sections — several independent diagrams, notes, and headings sitting side by side. With layout: "board" they are placed in the order you declare them, wrapping into rows like a poster. No edges required.
  • Or pin it exactly. With layout: "manual" every x/y (and a frame's width/height) is honoured verbatim — FigJam-style absolute placement — for a faithful, hand-arranged board.

1 · Auth & SSO — a request path, pinned

A single board that walks the team through a sign-in flow. A browser hits the app, which hands off to the identity provider; on success a session is minted, cached, and the user is let through. It doubles as a tour of the vocabulary: stacks for piles of records, free-floating asides, a dashed working frame, colour-coded connectors you can read by hue, and a speech-bubble callout — all from one spec, placed to the pixel with layout: "manual".

Loading canvas…

Because the board is a graph, an agent asking the MCP for this component gets the real structure — idp —depends-on→ callback, session —links→ Access granted — not a screenshot to interpret.

2 · From password auth to SSO — the whole story

Now the wider picture: several independent diagrams on one surface — password auth today, the SSO variant, the session service internals with its asides and a dashed cleanup note, a framing question, the target flow, and a closing line. None of these are one connected flow. This board uses layout: "board": each section is laid out on its own, then the sections are packed in the order they appear in the spec, wrapping into rows. You author reading order; the poster falls out.

Loading canvas…

The whole poster is one authored document. Reorder the sections in the spec and the poster re-flows; add a section and it slots in where you wrote it. That is the point of board layout — a knowledge base where the structure you write is the structure agents read and humans see.

3 · User concerns → suggestions

A discovery board: a wall of verbatim user concerns (sticky-style cards), a coloured Suggestions frame, and the flow they're about. The concern cards have no edges between them — they're just captured, and board layout arranges them in the order you wrote them.

Loading canvas…

On this page

Built withsuperlore