Drawio

drawio — From Text to Professional Diagrams

drawio-skill (Agents365-ai, MIT) converts a natural-language description into editable .drawio XML and exports it to PNG / SVG / PDF / JPG through the native draw.io desktop CLI — no MCP server, no background daemon. It can also turn an existing codebase into an auto-laid-out structure diagram. This skill is the jeo-skills wrapper: it documents when to reach for draw.io, how to install the CLI + skill (including as a plugin), and how to drive the plan → generate → self-check → iterate → export loop.

When to use this skill

• The user wants a polished, precise diagram — architecture, network topology, microservices, cloud (AWS/Azure/GCP), strict UML class/sequence, ER diagram, flowchart, mind map, org chart, ML/DL model
• The user wants to visualize an existing codebase — import graphs for Python / JS-TS / Go / Rust, or a Python class-inheritance hierarchy, with no manual coordinates
• The user needs real official vendor icons (AWS Lambda, Kubernetes pod, Cisco, BPMN, …) instead of guessed shape=mxgraph.* blank boxes
• The user wants AI/LLM brand logos (OpenAI, Claude, Gemini, Mistral, Llama, Ollama, LangChain, …) that draw.io ships none of
• The user wants the output exported to PNG/SVG/PDF and kept editable, optionally self-checked and refined over several rounds
• The user explicitly asks for draw.io / .drawio files

When not to use this skill

• The user wants a casual, hand-drawn / whiteboard look → use excalidraw-skill or tldraw-skill (sibling upstream skills)
• The user wants diagrams-as-code that live in git and render in Markdown → use mermaid (general) or plantuml (UML in CI)
• The user wants a slide deck / presentation artifact → presentation-builder
• The user only wants to browse/install Claude plugins generally → ccpi-marketplace
• The user needs AI image/video generation (not diagrams) → god-tibo-imagen / video-production

Prerequisites

Instructions

Step 1 — Install the draw.io desktop CLI

Verify with drawio --version. On WSL2 the CLI is the Windows desktop exe reached via /mnt/c — the skill detects this automatically.

Step 2 — Install the skill

# Any agent (Claude Code, Cursor, Copilot, …) — upstream bundle
npx skills add Agents365-ai/365-skills -g

# Claude Code plugin marketplace (upstream)
> /plugin marketplace add Agents365-ai/365-skills
> /plugin install drawio

# Manual install
git clone https://github.com/Agents365-ai/drawio-skill.git \
  ~/.claude/skills/drawio-skill

For the jeo-skills plugin install of this wrapper, see Step 8.

Step 3 — Generate a diagram from text

After installation, just describe what you want — the skill plans the layout, generates .drawio XML, exports a draft, self-checks, and iterates:

Create a microservices e-commerce architecture with Mobile/Web/Admin
clients, an API Gateway (auth + rate limiting + routing), Auth/User/Order/
Product/Payment services, a Kafka queue, a Notification service, and
User DB / Order DB / Product DB / Redis Cache / Stripe API.

Draw a Transformer encoder-decoder: 6-layer encoder with self-attention,
6-layer decoder with cross-attention, input embeddings (batch × 512 × 768),
positional encoding, final output projection. Annotate tensor shapes and
color-code by layer type.

Step 4 — Visualize an existing codebase

Turn code into structure diagrams with the bundled extract → auto-layout → validate pipeline (no manual coordinates):

# Import graph — Python / JS-TS / Go / Rust
python3 scripts/pyimports.py   myproject --group -o graph.json
python3 scripts/jsimports.py   ./src     --group -o graph.json
python3 scripts/goimports.py   ./module  --group -o graph.json
python3 scripts/rustimports.py ./crate   --group -o graph.json

# Python class-inheritance hierarchy
python3 scripts/pyclasses.py   mypackage --group -o graph.json

# any extractor → auto-layout → editable .drawio
python3 scripts/autolayout.py  graph.json -o diagram.drawio

Auto-layout uses Graphviz placement + orthogonal routing, transitive reduction (drops edges implied by a longer path), and --group for nested module containers. validate.py lints the .drawio (dangling edges, duplicate ids, overlaps) before the visual self-check.

Step 5 — Resolve real shapes and AI/LLM logos

# Search 10,000+ official AWS/Azure/GCP/Cisco/K8s/UML/BPMN shapes
python3 scripts/shapesearch.py "aws lambda" --limit 5

# Resolve one of 321 AI/LLM brand logos (lobe-icons, MIT)
python3 scripts/aiicons.py "claude" --json      # CDN-referenced (default)
python3 scripts/aiicons.py "openai" --embed     # self-contained data URI

Use these instead of guessing shape= strings, so vendor icons render correctly rather than falling back to a blank box.

Step 6 — Self-check, iterate, and style

The workflow runs check deps → plan layout → generate XML → export draft PNG → self-check + auto-fix (up to 2 rounds) → show user → 5-round feedback loop → final export. Built-in style presets are default, corporate, handdrawn, and the skill can learn a new style from a .drawio file or a flat image (saved only after you approve):

Draw a microservices architecture using my "corporate" style
Learn my style from ~/diagrams/brand.drawio as "mybrand"

Step 7 — Choose the right diagram type

Step 8 — Plugin-style installation alongside jeo-skills

This wrapper folder is plugin-installable through the standard jeo-skills flow so the routing guide, references, and installer land on disk for any supported agent runtime:

# Project install (writes into .agents/skills/drawio/)
npx skills add https://github.com/akillness/jeo-skills --skill drawio

# Global install for every detected agent
npx skills add -g https://github.com/akillness/jeo-skills --skill drawio

# Target specific agents
npx skills add -g https://github.com/akillness/jeo-skills --skill drawio -a claude-code -a codex -y

The skill also ships scripts/install.sh — a one-shot helper that detects/installs the draw.io CLI per platform and pulls the upstream drawio-skill bundle. See references/usage.md for the full command/flag reference.

Output format

When the user asks drawio for help, return a compact brief:

# drawio Routing Brief

## Scope
- Task: text-to-diagram | visualize-codebase | shape/logo-resolve | restyle
- Diagram type: architecture | uml | sequence | flowchart | erd | ml-dl | other
- Export target: png | svg | pdf | jpg | drawio-only

## Recommended next move
- install-cli | install-skill | generate | extract+autolayout | shapesearch | aiicons | refine

## Why
- 2-3 bullets grounded in the user's request

## Route-outs
- `mermaid` / `plantuml` for diagrams-as-code in git
- `excalidraw` / `tldraw` for hand-drawn/whiteboard looks
- `presentation-builder` for slide decks

Best practices

1. Confirm the CLI first — drawio --version before promising an export; without it you can only emit .drawio XML, not PNG/SVG/PDF.
2. Search shapes, don't guess — shapesearch.py resolves the exact official style string; guessed shape=mxgraph.* names render blank.
3. Use aiicons.py for LLM-app diagrams — draw.io ships no modern AI/LLM logos; --embed inlines them for offline rendering.
4. Let the self-check run — reading the rendered PNG and auto-fixing overlaps/clipped labels is the skill's main quality lever.
5. Reduce, then lay out — for codebases, transitive reduction turns a dense hairball into a traceable graph before auto-layout.
6. Pick the right family — reach for mermaid/plantuml for git-tracked, Markdown-embeddable diagrams; draw.io is for polished, exportable, icon-rich artifacts.

References

• Upstream repo: https://github.com/Agents365-ai/drawio-skill
• Online docs: https://agents365-ai.github.io/drawio-skill/
• Usage walkthrough: docs/USAGE.md
• CLI install recipes: docs/INSTALL_CLI.md
• Auto-layout reference: references/autolayout.md
• Shapes cheatsheet: references/shapes.md
• Troubleshooting (incl. WSL2): references/troubleshooting.md
• Local installer: scripts/install.sh
• Local usage/flag reference: references/usage.md
• Sibling diagram skills: excalidraw-skill, mermaid, plantuml, tldraw-skill
• Adjacent jeo-skills: ../presentation-builder/SKILL.md, ../cli-anything/SKILL.md, ../ccpi-marketplace/SKILL.md
• License: MIT (see upstream LICENSE)