Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Scientific Schematics

Describe a scientific diagram in plain language and get a publication-oriented PNG that an AI reviewer scores against your document type's quality bar.

Image & VideoIntermediate★ 33,030⑂ 3,248AI score 8/10Last updated: Aug 9, 2026

What it does

You describe a diagram; an image model (Nano Banana 2) draws it, and a reviewer model (Gemini 3.6 Flash) scores the result against a threshold that depends on where the figure will appear — 8.5 for journals, 8.0 for conferences and theses, 7.5 for preprints, 6.5 for slides. If the first pass clears the bar it stops; otherwise it rewrites the prompt from the critique and regenerates, up to two iterations.

Outputs are versioned PNGs (name_v1.png, name_v2.png), a copy at your requested path, and a name_review_log.json containing the score, critique, and stop reason. If the review call fails, no score is fabricated — the log records score: null and tells you to inspect the image yourself.

Stated limits: raster PNG only, no SVG/EPS/PDF vector output, no DPI or CMYK control, and image models routinely misspell labels, so manual proofreading is required.

Who it's for

  • Researchers who need concept, architecture, or pipeline figures for papers and theses
  • Clinical and meta-analysis authors who repeatedly draw CONSORT/PRISMA flowcharts
  • Grad students producing poster or slide graphics under deadline
  • Anyone who would rather not fight Illustrator or TikZ

Examples

  1. Trial flowchart: python scripts/generate_schematic.py "CONSORT participant flow with 500 screened, 150 excluded, 350 randomized" -o figures/consort.png --doc-type journal — iterates until it hits 8.5.
  2. Model architecture: "Transformer encoder-decoder, encoder stack left, decoder right, showing multi-head and cross-attention" with --doc-type presentation for a faster, lower-bar run.
  3. Signaling pathway: "EGFR → RAS → RAF → MEK → ERK → nucleus with phosphorylation steps labelled" for a poster, then read the critique field to catch any missing component before submitting.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/scientific-schematics folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/scientific-schematics/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/K-Dense-AI/scientific-agent-skills.git /tmp/sci-skills && mkdir -p ~/.claude/skills && cp -r /tmp/sci-skills/skills/scientific-schematics ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Clone the repository: git clone https://github.com/K-Dense-AI/scientific-agent-skills.git
  2. Copy the skill into your Claude Code skills folder: mkdir -p ~/.claude/skills && cp -r scientific-agent-skills/skills/scientific-schematics ~/.claude/skills/
  3. Install the Python dependency: pip install requests (or uv pip install requests)
  4. Create an API key at https://openrouter.ai/keys — image generation and review calls are billed per use
  5. Export the key: export OPENROUTER_API_KEY='sk-or-v1-...', and add the line to ~/.zshrc or ~/.bashrc to make it permanent
  6. Restart Claude Code and ask something like "make a CONSORT flow diagram for my journal paper"
  7. Open the PNG in figures/ plus the matching *_review_log.json to verify the score and proofread every label
  8. Privacy note: your prompt and the generated image are sent to OpenRouter, so keep unpublished data, patient details, and embargoed material out of the description.
View source on GitHub ↗License: MIT