Table Image Skill

Render markdown tables and diagrams as hand-drawn sketch images using Gemini's image generation.

Workflow

1. Receive Table Content

Get the markdown table from the user. The table should be in standard markdown format.

2. Construct the Prompt

Build a prompt using this structure:

Black and white hand-drawn sketch, loose ink strokes. A clean technical table drawn on
white paper. Minimalist, high contrast, editorial tech illustration. Compact layout,
minimal margins, cropped tightly to the table edges, the table fills the frame. --ar 24:9

[paste the full markdown table here]

3. Generate

uv run ~/.claude/skills/_generate_image.py "<constructed prompt>" --ar 24:9 -o table.png

4. Report

Report the saved file path and suggest reviewing the image.

Example

Input table:

| Feature | Go | Python | Ruby |
|---------|-----|--------|------|
| Concurrency | goroutines | asyncio | threads |
| Typing | static | dynamic | dynamic |
| Speed | fast | moderate | moderate |

Prompt:

Black and white hand-drawn sketch, loose ink strokes. A clean technical table drawn on
white paper. Minimalist, high contrast, editorial tech illustration. Compact layout,
minimal margins, cropped tightly to the table edges, the table fills the frame. --ar 24:9

| Feature | Go | Python | Ruby |
|---------|-----|--------|------|
| Concurrency | goroutines | asyncio | threads |
| Typing | static | dynamic | dynamic |
| Speed | fast | moderate | moderate |

Notes

• Requires GEMINI_API_KEY or GOOGLE_API_KEY in environment
• Default output: table.png in current directory
• Default aspect ratio: 24:9 (wide, fills frame with table)
• Uses shared script: ~/.claude/skills/_generate_image.py
• For complex tables, results improve when the table has clear headers and consistent column widths