drawio-ai-kit provides a zero-runtime-dependency CLI to orchestrate and validate draw.io diagrams with domain skills for AWS, Azure, GCP, Databricks, and BPMN. It includes a declarative catalog, design principles, and a structural validator with vision self-check capabilities.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (2 so far).
What it is
An orchestration and validation framework enabling AI agents to generate structurally precise and aesthetically standardized draw.io diagrams, optimized for AWS, Azure & GCP architectures. Exposed to the AI via the zero-dependency drawio-ai CLI. The kit includes a Declarative Catalog, Design Principles, and a Structural Validator to audit diagram XML before serialization.
How it works
The tool defines a diagram topology and nested structure; a layout engine computes coordinates and sizes, while a validator checks stencil references, nesting, colors, and geometry. It supports a render/vision self-check workflow via the drawio-ai CLI and domain-specific Skill integration. Commands provided include search, style, validate, audit, render, categories, principles, and root among others. Domain Skills wrap these commands into a full build→validate→render→write workflow. The architecture relies on rules, a catalog mapping stencil IDs to colors and points, and an engine that lays out diagrams programmatically.
Getting started
Full install example (CLI + 5 Domain Skills):
npm i -g github:sparklabx/drawio-ai-kit && npx skills add sparklabx/drawio-ai-kit
Restart agent, then try: "draw an AWS 3-tier web app". The first command installs the drawio-ai binary; the second registers Domain Skills. Optional guidance includes using --skill drawio-aws to add a single domain and details in INSTALL.md for pinning versions or installing from a clone.
In addition, installation and update tips are provided:
npm uninstall -g drawio-ai-kit
npx skills remove drawio-aws
npm i -g github:sparklabx/drawio-ai-kit
npx skills update
To render with vision self-check:
drawio-ai render out.drawio -o out.png
Commands like drawio-ai search, drawio-ai style, drawio-ai validate, and drawio-ai render are highlighted in the README. The kit relies on Python and Node tooling as part of its runtime assets.
Recent releases
- v1.0.1 (2026-07-15): search & workflow fixes; patch release with bug fixes and hardening; no breaking changes.
- v1.0.0 (2026-07-10): CLI-only surface; first major release; zero runtime dependencies; five thin Domain Skills.
- v0.3.0 (2026-07-03): Multi-cloud support with Azure, GCP, Databricks; authentication per cloud; expanded example library.
- v0.2.0 (2026-06-27): Additions under What’s new; multi-agent support via Antigravity CLI and Gemini CLI features.
Traction
Stars: 595 (as of latest data). Forks: 103. Open issues: 1. Language: JavaScript. License: MIT. Created: 2026-06-17. Last push: 2026-07-18.
Behind the repo
The project ships 5 Domain Skills: drawio-aws, drawio-azure, drawio-gcp, drawio-databricks, drawio-bpmn, each wrapping the main CLI into domain-specific workflows. The core tooling runs through the drawio-ai CLI and uses vendored helpers in vendor/ such as autolayout.py, aiicons.py, repair_png.py, and encode_drawio_url.py.
Caveats
License: MIT. Runtime ecosystem: Node 18+ required; Python 3.11 for data ingestion and compilation pipeline. The README notes zero runtime dependencies and no postinstall hooks. The Release notes show breaking changes at v1.0.0 (MCP server removed).