Source: https://github.com/aipoch/medical-research-skills

When to Use

You need scientific-looking diagrams/posters (laboratory poster aesthetic) generated from a short natural-language brief.
The user requests SVG output specifically (e.g., “output SVG”, “vector graphic”, “embeddable in a web page”).
You want language-to-image results without using diffusion/LLM image models, prioritizing interpretable structure over photorealism.
You need repeatable, parameter-controlled visuals (seed/palette/structure) for research notes, slides, or documentation.
You want a structured visualization (grids, networks, waveforms, symbol rings) rather than an illustrative drawing.

Key Features

Converts a natural-language brief into a renderable SVG with a scientific, restrained visual style.
Multiple built-in styles via STYLE:
- lab-atlas (default): calm, stable, laboratory map feel
- signal-loom: denser spectral waveforms, stronger texture
- lattice-field: prominent lattice grids, denser nodes
Produces SVG + JSON metadata (e.g., prompt, seed, palette) for traceability.
Writes a convenience preview file: output/svggen/latest.svg.
Tunable density and composition controls (e.g., nodes, noise, bands, rings).