canvas_read

TOOLtools/call

Read a canvas as nodes (id, shape, text, x, y, w, h, frame, link, color), edges (id, from, to, label, style, heads) and frames, with the canvas seq. With since: a seq you read before, only what changed after it, plus removed ids. With format mermaid: the board as a Mermaid flowchart. Pass the seq you read as baseSeq when you write, so you never overwrite a person's later change.

Scopes

read:projects

A project canvas also needs its project to be readable, and for writes writable (Edit tasks, not archived) by the key's person. A workspace canvas is readable when its person owns it or it is shared with the workspace, and writable when its person owns it or holds Edit tasks on a shared one.

Inputs

canvasIdstringRequired
sinceinteger: Only changes after this seq.
format"json" | "mermaid"
cursorstring

Returns

The canvas as complete agent JSON for a bounded page (or whole-board Mermaid text), with its seq and next cursor. Page size follows native read budgets; follow nextCursor until null. Every internal context continuation checks the original live access and the same seq; sequence drift refuses the read. A call returns content[0].text with the JSON result and structuredContent.result with the same value.

Listed when

The key has read:projects and Canvas is open for the workspace; the tools that write need write:projects too.