Why AI-Ready?

dartwork-mpl is designed from the ground up to work with AI coding assistants. This page shows the concrete features that make AI-generated plots reliable.

Design rationale: For the philosophy behind these decisions — why thin utilities beat wrappers for AI — see Designed for AI Agents.

Automatic Validation

AI-generated plots often have subtle issues invisible in text-only environments: clipped labels, overlapping legends, missing tick marks. dartwork-mpl catches these automatically:

# After creating your figure
issues = dm.validate_figure(fig)

# Returns a list of detected problems:
# - Clipped text outside figure bounds
# - Overlapping elements
# - Inconsistent font sizes

This is especially powerful in autonomous AI pipelines where there’s no human to visually inspect every plot. The validation step acts as a quality gate.

MCP Protocol: Real-Time AI Access

The Model Context Protocol (MCP) gives AI assistants live access to dartwork-mpl’s documentation, colors, styles, and validation tools:

Category

Count

Examples

Resources

8

Usage guides, color palettes, font lists, style presets, plot templates

Tools

7

Color lookup, code linting, data validation, GitHub file fetch

Prompts

2

Guided plot creation, style compliance review

The AI assistant always has the latest, most accurate documentation — no copy-paste, no stale caches, no hallucinated APIs. Tools like lint_dartwork_mpl_code catch antipatterns (e.g., figsize=, tight_layout()) automatically, and validate_plot_data ensures data structures are correct before plotting.

→ For the full list and setup instructions, see MCP Server.

Putting It All Together

🗣️ User Prompt

You say: “Plot the signal response with a red line and save it as SVG for my paper.”

↓

🧠 Context Retrieval (MCP)

AI reads dartwork-mpl guide via MCP:

  • Knows to use dm.style.use('scientific')

  • Knows to use dm.simple_layout(), not tight_layout

  • Knows color syntax: 'oc.red5'

  • Knows to save with dm.save_formats()

↓

✅ Execution & Validation

  • ✨ AI generates correct code on first attempt

  • 🔍 dm.validate_figure() confirms no issues

  • 📄 Paper-ready SVG saved

→ Ready to set this up? See AI-Assisted Development for the workflow guide, or jump straight to MCP Server for configuration.