scientific-schematics

To generate publication-ready scientific diagrams, provide a natural-language description of your visual to the generator script, specifying the output path and document type. The Skill creates publication-quality images for research communication, such as neural network architectures or biological pathways.

Iterative refinement automatically improves diagram quality by running an image generation step followed by a vision-based quality review. If the diagram score falls below a document-type threshold, it regenerates the image and outputs versioned images with JSON review logs.

You can create scientific diagrams for research communication including neural network architectures, biological pathways, CONSORT or PRISMA-style flowcharts, circuit schematics, and system or block diagrams from natural-language descriptions.

The diagram generator applies document-type-aware quality thresholds, enforcing stricter requirements for journal articles than for presentations. This ensures the output meets the specific readability and scientific accuracy standards required for each publication format.

Accessibility and scientific accuracy are handled through colorblind-friendly styling and scientific accuracy checks during the vision-based quality review. The generator outputs versioned images and captures detailed JSON review logs per iteration to verify publication standards.

You need the requests dependency and an OpenRouter API connection to run the scientific diagram generation script. The script uses OpenRouter with Gemini 3.1 Pro Preview to perform the vision-based quality review and threshold-based regeneration.