
from drawio-skill360
Generate professional draw.io (.drawio) diagrams from natural language and export to PNG/SVG/PDF using the draw.io desktop CLI.
Generates editable .drawio XML diagrams from plain-language prompts, exports previews to PNG/SVG/PDF/JPG via the draw.io desktop CLI, and supports an iterative review loop for edits. It can embed the diagram XML into exported PNG/SVG/PDF so files remain editable. Includes presets for flowcharts, ERD, UML, sequence diagrams, architecture, and ML model diagrams.
Use when a user asks for diagrams, flowcharts, architecture drawings, ER/UML diagrams, or visualizations of systems with 3+ components. Also trigger proactively when explaining complex data flows or multi-step processes that benefit from a visual representation. Skip for very small lists or quick Q&A.
Best used with agents that can run local commands and (optionally) image-vision self-checks (e.g., Claude Sonnet/Opus, OpenClaw agents that expose CLI execution).
Draw.io diagram generation skill with self-check loop, 6 diagram presets, style presets system, and iterative refinement. Two bundled scripts (repair_png.py for fixing truncated IEND in -e PNGs, encode_drawio_url.py for browser fallback URLs) are clean, minimal stdlib Python. SKILL.md is exceptionally well-structured with progressive disclosure via references/ and clear workflow steps. Auto-update check asks user consent before git pull. No security concerns.
Exemplary skill structure — lean SKILL.md with on-demand references, clean scripts, thorough troubleshooting table. Auto-update mechanism throttled to 24h and requires explicit user consent. Style presets system is well-designed with proper validation and fallback ladder. Minor: skill_md_body was null in DB fetch, had to clone repo for full analysis.