Verified project record

Agents365-ai/drawio-skill

drawio-skill is an Agent Skills-compatible skill that turns natural language, code, Terraform/Kubernetes, SQL, OpenAPI and other sources into editable .drawio architecture diagrams, with incremental sync, drift diff and CI architecture tests. It ships as a single SKILL.md, portable to any Agent Skills-compatible agent, with an optional MCP server exposing the diagramctl CLI workflows.

Project overview

It addresses the common problem of architecture diagrams going stale as systems change by generating maintainable, tested .drawio files directly from real sources across 16 extractors, including Python/JS-TS/Go/Rust code, Terraform/Kubernetes/docker-compose, SQL DDL, OpenAPI, AsyncAPI, Protobuf, GraphQL and CI configs.

Project type: Workflow · Prompt Engineering
Use cases: Documents & Office · Automation
Deployment: Refer to project documentation
License: MIT

Best for

Developers and AI engineers who want maintainable .drawio architecture diagrams generated from natural language or real code and infrastructure sources.
Users of Agent Skills-compatible agents, since the skill is a single SKILL.md requiring no MCP server.

Key capabilities

Plans layout, writes .drawio XML, exports, self-checks its own PNG and auto-fixes overlaps, clipped labels and stacked edges.
16 extractors turn Python/JS-TS/Go/Rust code, Terraform/Kubernetes/docker-compose, SQL DDL, OpenAPI, AsyncAPI, Protobuf, GraphQL and CI configs into auto-laid-out editable .drawio diagrams with official cloud icons.
Single CLI covering build/sync/views/query/test/review/whatif/story/publish/transform; core workflows are stdlib-only and offline, and an optional MCP server exposes them to MCP hosts.

Limitations and risks

Mermaid to .drawio conversion and the ELK --layout pass require draw.io CLI version 30 or higher.
AI/LLM logo icons may be fetched from the unpkg CDN at render time; --embed inlines them for offline use.

Getting started

Install the draw.io desktop CLI (e.g., brew install --cask drawio or a platform installer) and verify with drawio --version; then run npx skills add Agents365-ai/drawio-skill -g; then describe the diagram you want in a compatible agent. Setup difficulty is medium because it requires installing the draw.io desktop CLI and then installing the skill into an agent host. No coding is required.

Evidence and sources

GitHub project description: Agent skill that turns natural language, code, Terraform/K8s, SQL, OpenAPI, AsyncAPI, Protobuf and GraphQL sources into editable, tested draw.io architecture diagrams
README: **One CLI, optional MCP server** — `diagramctl doctor/build/sync/views/query/test/review/whatif/story/publish/transform`, core workflows stdlib-only and offline; the MCP server ex…
README: **Describe it, get an editable `.drawio`** — the skill plans the layout, writes the XML, exports, then self-checks its own PNG and auto-fixes overlaps, clipped labels, and stacked…
README: 16 extractors: **Python · JS/TS · Go · Rust** import graphs, **Python class inheritance**, **Terraform / Kubernetes / docker-compose** with official cloud icons, **live** infra fr…
README: Portable to any Agent Skills-compatible agent, no daemon

Source updated at: 2026-10-12T19:49:25Z

Open original GitHub project