Plot example — a cell-size histogram

A complete plot — the two files a plot is, in the two-file layout the Plots tab expects: plot.yaml (the manifest: what port it binds to, how it renders) and render.py (the script: render(collection) → a matplotlib figure). Copy the folder into your project's plots/ directory and relink the target in the GUI, or scaffold a fresh plot with the scaffold_plot tool and compare.

What to notice

• Binding is by node id and port, never by label. The target block names the workflow file, the node_id, and the output_port; two nodes with the same display label stay distinct.
• The script imports nothing from SciStudio. Everything arrives through collection: collection.items.open() reads every table in the batch, open_one() the first. (A package type opens as its core base type — an Image opens as a NumPy array.)
• render(collection) — exactly that signature. Returning None is an error; return a figure, a path to an image you wrote in the working directory, or a list of either.
• A plot is preview-only. It never becomes a workflow node and never feeds another step; the rendered files are a cache the next run overwrites. A figure that must be kept is a block output (an Artifact), not a plot.

The script pools one column across every table in the batch into a single histogram — the smallest realistic render. For the full contract (house style, figure sizes, the R/ggplot2 form), see the writing-plots.md page of this guide.