Edit the figure. Keep the code.

A visual editor for matplotlib and AI-generated scientific figures. Click what you want to change, arrange the panels in millimetres, check publication requirements, and export — without losing the code behind the figure.

Download Tavotto View on GitHub

Or from PyPI, on any platform: pipx install "tavotto[worker]"

150 × 112.5 mm

Your plot is finished. Figure 1 isn’t.

The last stretch before submission goes: change a line, re-run the script, look, change it again. Twenty times, for things you can see but not easily say — the legend three millimetres to the left, the tick labels one point smaller, panel (b) aligned to panel (a).

The other way out is to finish the PDFs by hand in a vector editor, and accept that the figure and the code that made it have parted company.

Tavotto is the third option. Open the figure, change what you can see, export. The script stays where it was, and every change can be undone.

The old loop

1 change code
2 run script
3 inspect
4 change code
5 run again
6 repeat

With Tavotto

1 click the title
2 drag the legend
3 align the panels
4 export

matplotlib draws the plot. Tavotto is the visual last mile — it does not replace your plotting code.

From plot to Figure 1

Close-up of figure editing in Tavotto: the title “Reaction kinetics” selected on the canvas, and the properties panel showing its font, size of 9 pt and the source script fig1_kinetics.py.

01 / Edit

Edit what matplotlib already drew.

Double-click a panel and Tavotto runs your script once, keeping the matplotlib Figure in memory. From then on you are editing that figure directly: pick the title, an axis label, a tick, a curve, the legend, an arrow your script drew — from the element tree or by clicking it on the canvas — and change its size, colour, weight, dash pattern, or drag it somewhere else. The figure follows the cursor frame by frame; matplotlib runs once, when you let go.

Title · legend · curves
Axes, ticks and their formatters
Fonts, colours, positions
Arrows and annotations

Serif · 9 pt · fig1_kinetics.py

02 / Arrange

Build Figure 1 in millimetres.

Panels land on a page measured in millimetres, at the size their script drew them. Drag them, snap them to each other, align and distribute a selection, or bind panels into a row, column or grid that reflows when a size changes. Panel labels (a)(b)(c) come from one command, and annotations — text, arrows, shapes — sit on top.

Physical page, physical sizes
Snap, align, distribute, group
Rows, columns and grids that reflow
Panel labels from one command

Tavotto in layout mode: the asset library on the left with three source PDFs and their physical sizes, the composed page in the middle with one panel selected, and its position, millimetre size and scale on the right.

03 / Check

Catch publication problems before the reviewer does.

Before anything is written, Tavotto checks the figure against a publication profile. Font sizes are judged at their final physical size — a panel placed at 60% is checked against fontsize × 0.6, not against what the script asked for.

Blocking: Stops the export until you confirm in writing.
Warning: Always shown, never silently dropped.
Not verifiable: What genuinely cannot be checked — text inside an imported bitmap — and needs a human.
Suggestion: Never decides anything for you.

Checks run against a publication profile you can override per journal. Tavotto flags what can be checked — it does not promise acceptance.

04 / Export

Export the figure, not a screenshot of it.

PDF export embeds each original vector panel as it was drawn, so the text stays real, selectable, searchable text. PNG is rasterised from that same PDF, so the two can never disagree. The preflight findings — and anything you confirmed past — can be written into a proof report beside the exported files.

Vector PDF, text stays text
PNG rasterised from the same PDF
Optional proof report on the side

tavottofile/export/

Fig1.pdf vector · real text
Fig1.png 300 dpi · same pdf
Fig1_proof.json findings + confirmations

VECTOR PDF · 300 DPI PNG

05 / Workflow

Let AI write the first draft. Finish the figure visually.

Coding agents write good first-pass matplotlib. What is left over is visual — the legend two lines too tall, the panel that wants to be a little smaller, the label overlapping a data point. Describing that in a prompt is slower than doing it.

Workflow: a Python script — written by you, Claude or Codex — runs matplotlib, which produces a vector PDF; Tavotto handles visual editing, layout and the publication preflight; the result is a PDF or PNG. The script stays the source throughout.

Hand a figure over with one command

tavotto open figures/Fig1_kinetics.pdf

From a terminal or from an agent. It opens that figure’s library as a project and hands the figure to the app — into the window that is already open, if there is one.

Codex plugin

A skill, a local MCP server, and an embedded canvas built from the same frontend code the desktop app runs. Parts of the Codex Desktop integration are not yet verified inside a real Codex Desktop — the plugin README says which.

The script stays the source throughout.

06 / Publication

Built for the last 5% before submission.

Publication rules are physical: column widths in millimetres, font floors in points, raster floors in dpi. Tavotto measures the finished page against them.

80 / 150 mm: Single- and double-column widths in the default profile
8.5 pt: Strict font floor — judged at final physical size, 8 pt absolute
300 dpi: Minimum for rasterised output
0.5–1.5 pt: Line-width steps the frame and curves are checked against

The rules live in one versioned profile that engine and interface both read. A journal with its own widths needs a one-line override, not a fork — and the override is recorded in the proof report.

07 / Local-first

Your figures stay on your machine.

Rendering, composition and export are local processes. Tavotto does not upload your figures, scripts, project files or data anywhere — unpublished results do not leave the building.

Local rendering
Local layout
Local export
No figure upload
No source upload

Exactly two requests of its own

A once-a-day check of GitHub Releases for a new version, and — only after you explicitly opt in — anonymous usage statistics whose event schema cannot represent filenames, paths, figure text or prompts. Each has its own off switch; neither covers the other.

The optional AI assistant hands a request to a Codex or Claude CLI on your machine, only when you ask it to.

Read the privacy policy

Finish the figure without rewriting it.