Genops diagrams

Skill OmerMohideen/genops/skills/genops-diagrams

Author validated draw.io (.drawio) cloud architecture diagrams using official AWS/Azure/GCP icon styles. Use when asked to draw, diagram, or visualize a cloud architecture/deployment, generate a draw.io/diagrams.net file, or produce an architecture diagram for infrastructure (works well alongside the genops IaC skill).From its SKILL.md

Install

npx -y skills add OmerMohideen/genops --skill genops-diagrams

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.9 KB, as published. Nobody here has run it

GenOps Diagrams

Turn an architecture description into a validated .drawio diagram, delivered as a diagrams.net URL (default) or a saved .drawio file.

Workflow

Resolve the cloud provider. If ambiguous, use the provider named in the request.
Plan the diagram: pick services, group them into containers (cloud, region/account/project, VPC/VNet, subnets), decide layout direction (left-to-right: clients → edge → compute → data).
Look up every service icon — never invent or transcribe a shape=mxgraph.* style by hand:
```
python scripts/drawio_tools.py search <aws|azure|gcp|generic> <keywords>
```
Reference the result by its alias in the cell style: style="icon:<provider>:<key>" for a service, style="group:<provider>:<key>" for a container. Use the catalog's width/height for the mxGeometry.
Author the XML following references/diagram-rules.md (file skeleton, vertices, containers, edges, color palette, layout/spacing rules).
Validate, fixing every error (re-validate after fixes, at most two rounds):
```
python scripts/drawio_tools.py validate <file|->
```
Deliver:
- Default — no target directory given: build a shareable URL that opens directly in the browser:
```
python scripts/drawio_tools.py url <file|->
```
  Reply using EXACTLY this markdown link syntax: [Open diagram](<url>) — square brackets around the text, the url in parentheses immediately after, no space between ] and (. Do NOT write text (url), do NOT write the url in a code block, do NOT paste the raw url as visible text anywhere in the reply — all several hundred characters of percent-encoding must be hidden behind the link text.
- Only if the user gave an explicit target directory: save the file instead:
```
python scripts/drawio_tools.py save <file|-> <target_directory> [--filename NAME] [--overwrite]
```
  (save and url both validate internally — a separate validate call first is for iterating on errors, not required before them.)

<file|-> accepts a path to the XML or - to read it from stdin.

Rules

Icons by alias only. Unknown aliases fail validation with the exact valid keys for that provider/kind.
Every edge needs an <mxGeometry relative="1" as="geometry" /> child — a self-closing edge cell does not render.
Snap every x/y/width/height to multiples of 10; keep ≥20px between sibling cells and ≥30px between a container's children and its border.
Label every element with the official service name (Amazon EC2, AWS Lambda, Azure App Service, Cloud Run). No decorative extras.
Scope: cloud architecture diagrams only.

Bundled icon catalogs (`assets/icons/`)

aws.json, azure.json, gcp.json, generic.json — official draw.io stencil aliases, one JSON per provider. Each entry has key, name, category, kind (icon or group), alias, width, height. Search them with the search subcommand rather than reading the files directly.

Coverage: AWS 53, Azure 37, GCP 15, generic 5 icons (plus a handful of container group: aliases per provider), verified against draw.io's bundled aws4/azure2 stencil sources. GCP's product icons are hand-embedded full-color SVGs rather than named stencils, so its catalog is thinner — search returning no matches, or validate erroring Unknown icon alias, means fall back to the plain shapes in references/diagram-rules.md (rounded=1, shape=cylinder3, shape=cloud) rather than guessing a shape=mxgraph.* string.