Make a diagram readable with AI
Fit an auto-generated graph to the space where it will actually appear, then check that its labels and connections still tell the same story.
Make a diagram readable with AI by measuring it where the reader will see it. A graph that looks perfectly clear when opened in its own tab can shrink to a grey strip in a documentation column. The restore-diagram-legibility skill works from the diagram source and the rendered dimensions to discover which part is consuming the space.
Try changing the longest labels before adjusting the whole graph. A short line break can reduce the width of a node. Disconnected components may need a different arrangement, and a dense graph may work better as several related views. Check each change at the target width and preserve meaningful relationships.
What it does
- Measure an important label in the final display context.
- Wrap a long label in the editable source.
- Compare horizontal and vertical layouts using rendered output.
- Separate clusters while preserving the explanation of their connections.
- Check that layout-only links cannot be mistaken for genuine dependencies.
Install and use
- Install the skill in your agent. Provide editable Mermaid or DOT source, a compatible renderer and the width available on the page.
- Ask for a before-and-after comparison of the same graph in a narrow documentation column.
- Inspect the chosen version, verify its edges against the original and record any labels still too small to read.
FAQ
Source
Made by Philipp Thoss. Original skill: restore-diagram-legibility, published on GitHub under the MIT licence.
View on GitHub