CodeBook Tour

Think in narrative, compute in place. The CodeBook is Circuitry's notebook surface: a single scrolling document made of cells — code, prose, charts, sheets, images, hand-drawn figures, and AI agents — that all run against one persistent Python state. It's the best surface for work where the reasoning matters as much as the result: analyses you'll re-run, explorations you'll explain, experiments where each step builds on the last.

And here's the part that makes it more than a notebook: a CodeBook is a workflow. Every notebook you write is simultaneously a graph of connected nodes, and any workflow can be read as a notebook. Sketch a pipeline on the workflow canvas, then flip it into notebook form and narrate it cell by cell — or start from prose and code, and flip it into the graph view when you want to see the plumbing. One document, two ways of thinking about it.

Cells for every kind of thinking

A brand-new CodeBook greets you with Add Code and Add Text buttons. After that, hover between any two cells (or below the last one) and a small Add Cell pill appears; hovering expands it into the full cell palette:

• Code — a code cell: write and run code against the notebook's persistent state.
• Text — a text cell: prose with a rich-formatting toolbar.
• Agent — an agent cell: an AI step that lives inline in the document.
• Sheet — a sheet cell: an editable data grid inside the notebook.
• Chart — a chart cell: visualize results in place.
• Image — an image cell.
• Figure — a figure cell: draw a diagram by hand, right in the flow of the document.

A tour of the controls

The cell toolbar

Focus any cell and a compact toolbar floats above it, with a drag handle and a type badge on the left (code cells also show their language):

• Drag to reorder — grab the handle and drag the cell anywhere in the document.
• Run Cell Run Cell — executes this cell. ⇧⏎ does the same from the keyboard.
• Formatting toolbar — on text cells, shows or hides the rich-formatting controls.
• Figure label — on figure cells, shows or hides the caption label.
• Edit figure — on figure cells, enters in-place drawing mode.
• Pair side-by-side — sets this cell next to the previous one in a two-column layout (press again to return to full width). Great for code-beside-chart or prose-beside-figure spreads.
• Cell Settings — opens this cell's configuration.
• Delete Cell — removes the cell (asks for confirmation first).

The cell gutter

The left margin of each cell carries quiet, always-there controls:

• / Fold code — the triangle collapses a code cell down to its output; folded cells show a thin Show code strip you can click to expand.
• Run — hover the gutter of an idle runnable cell and the execution badge becomes a run button.
• Figure captions — double-click a figure's caption to edit it inline.

The bottom toolbar

A floating pill sits centered at the bottom of the notebook:

• Show REPL — opens a console at the bottom of the notebook that shares the notebook's persistent Python state: poke at variables your cells just created, try a line before committing it to the document.
• Run All Cells Run All Cells — executes every cell in order, top to bottom.
• Reset — clears all cell outputs and restarts the Python state for a clean slate.
• Save Workflow Save — saves the notebook (and, because they're the same document, the workflow behind it).
• Show Config — opens the notebook's settings, including where its code runs.
• Reading mode — strips away all editing chrome for a clean, paper-like read; press to return to editing. Perfect for presenting results or sharing a finished analysis.
• Back to Workflow View Workflow View — flips the same document back to the graph view on the canvas.

The REPL

With the REPL open, every line you run becomes a transcript entry with hover actions:

• Clear History Clear History — empties the transcript.
• Copy — copies that entry's code.
• Promote to Cell — the experiment worked? Turn the REPL entry into a real notebook cell in one click.
• Delete — removes the entry from the transcript.
• Debug — on an entry that errored, hands the failure to AI to debug.

Keyboard

Works with the rest of the studio

The CodeBook is one surface of one studio — the same document is at home everywhere:

• Workflows — every CodeBook is a workflow; flip between notebook and graph views freely.
• Coming from Jupyter — how CodeBook concepts map to the notebooks you already know.
• Code nodes — the same code execution that powers cells, as a node on the canvas.
• Sheets — the full spreadsheet editor behind sheet cells.
• The agentic IDE — when an exploration graduates into a project, the coding agent picks it up.
• Sharing & Live Collaboration — share a finished notebook or work it together, live.