enterprise-svg-architect
Skill2 starsChanged 2 months agoMITView on GitHub
Draw enterprise-grade architecture and topology SVG diagrams — especially for identity/authentication, zero-trust and clientless (无端) access, SSO integration, reverse-proxy/gateway convergence, and system-integration flows. Produces self-contained SVGs with trust-boundary zones (public internet / DMZ / internal), component nodes, labeled data-flow arrows (protocols like HTTPS/302/SAML/OIDC/SSO), and a legend; auto-adapts to light/dark mode and exports to 2x PNG via a bundled script. Use when the user asks to draw or visualize a system architecture, network topology, deployment diagram, integration or 接入 diagram, an auth/login flow at the architectural level, trust boundaries, or how system X integrates into system Y. Trigger phrases — 架构图, 拓扑图, 部署图, 集成图, 接入图, 信任边界, 最终效果图, enterprise architecture diagram, topology diagram, zero-trust access diagram, SSO integration diagram.
In this repository6 other files it gives its agents.
- ameureka-spec-devSkill
- blindspot-auditSkill
- cowork-syncSkill
- fpf-reasoningSkill
- loop-prompt-generatorSkill
- requirements-matrix-generatorSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Do the layout math BEFORE writing SVG: Read references/svg-conventions.md for the coordinate system, CJK text-width calibration, box-sizing …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Enterprise SVG Architect
Overview
Produce precise, professional, self-contained architecture/topology SVGs that render identically inline (visualize plugin), in a browser, and as exported PNG. Diagrams encode meaning through trust-boundary zones, component nodes, and labeled data-flow arrows — not decoration. Every SVG is written to a file, adapts to light/dark mode via an embedded @media block, and is exported to 2x PNG for docs/PPT.
This skill is self-contained. It does NOT depend on the visualize plugin's pre-built CSS classes (c-blue, t, ts) — those exist only inside that host. Instead it defines its own <style> with concrete palette hex + dark-mode overrides, so the same SVG works everywhere.
Workflow
- Scope the diagram. Confirm three things (ask only if unclear): (a) the systems/components involved, (b) the trust zones each lives in (public / DMZ / internal / partner), (c) the primary flow(s) to show. Count the nouns — if >8 components, split into an overview + per-flow detail diagrams rather than one dense canvas.
- Pick a topology pattern. See
references/architecture-patterns.mdfor the standard enterprise layouts (zoned-lanes, reverse-proxy convergence, zero-trust access, SSO/identity federation, hub-and-spoke). Choose one; don't invent layout from scratch. - Do the layout math BEFORE writing SVG. Read
references/svg-conventions.mdfor the coordinate system, CJK text-width calibration, box-sizing formula, zone-nesting rules, and the mandatory viewBox checklist. Cramped/overlapping boxes are the #1 failure — compute widths and gaps first. - Start from the template. Copy
assets/template.svg— it already contains the light+dark<style>, the arrow marker, and a zone/node/flow example. Fill in nodes and flows. - Draw zones → components → flows → legend, in that z-order. Zones are large dashed containers (one ramp each). Components are solid nodes colored by category. Flows are arrows with short protocol labels placed in clear space. Add a one-line legend whenever color or line-style encodes meaning.
- Verify against the checklist at the bottom of
references/svg-conventions.md(viewBox fits, no overlaps, no arrow crosses an unrelated box, every text has a class, CJK labels fit their boxes, dark-mode overrides present for every color). - Write the
.svgfile, then export PNG withscripts/svg_to_png.py <file.svg>(defaults to 2x). Read the PNG back to visually confirm CJK rendered (no tofu boxes) and nothing clipped. - Render inline (if the visualize plugin is available) by passing the SVG source to
show_widget, and/or deliver the.svg+@2x.pngfiles.
Core rules (essentials — full detail in references)
- Canvas:
viewBox="0 0 680 H",width="100%",role="img"with<title>+<desc>first. Keep width 680 (matches inline host 1:1). Content stays in x=40..640. Set H = lowest element + ~24px. - Self-theming: define colors in a
<style>block using concrete palette hex, plus a@media (prefers-color-scheme: dark)override for EVERY color class. Never rely on host classes. Palette hex table is inreferences/svg-conventions.md. - Color encodes category, not sequence. ≤2 meaningful ramps + gray for neutral/structural. Zones and node-categories each get one ramp. Add a legend when color carries meaning.
- CJK is wide: ~15px per character at 14px, ~13px at 12px (Latin is ~8px/7px). Size every box from its longest label using the formula in the conventions file, or Chinese labels WILL overflow.
- Arrows: 1.5px, single-direction flow where possible; never route a line through an unrelated box (use an L-bend
<path>withfill="none"). Protocol labels (HTTPS, 302, SAML, OIDC) go in clear space, ≤3 words. - Trust boundaries: draw as large dashed rounded rects with a top-left label; nest components inside with ≥20px padding. Public internet, DMZ, and internal network are conventionally distinct zones — see patterns file.
- No gradients, shadows, blur, emoji, or
<style>color without a dark-mode counterpart. Two font sizes only: 14px (labels), 12px (subtitles/flow labels). Sentence case.
Bundled resources
references/svg-conventions.md— coordinate system, full palette hex table (9 ramps, light+dark), text/box sizing math with CJK calibration, arrow marker, zone-nesting rules, and the pre-flight verification checklist. Read before drawing.references/architecture-patterns.md— the enterprise topology patterns (zoned-lanes, reverse-proxy convergence, zero-trust/clientless access, SSO/identity federation, deployment/hub-spoke), each with when-to-use, a layout sketch, and node/flow conventions. Read to pick a layout.assets/template.svg— copy-to-start skeleton: light+dark<style>, arrow marker defs, one zone + one node + one flow + legend. Replace the example contents.scripts/svg_to_png.py— export SVG → PNG. Usage:python3 scripts/svg_to_png.py input.svg [-s SCALE] [-o out.png]. Default scale 2 (→input@2x.png). Usesrsvg-convert, falls back to headless Chrome; renders light mode (ignores@media dark), which is what you want for a shareable PNG. Run--helpfor options.
Delivery
Save both the .svg (adaptive source) and @2x.png (portable) next to the user's related files, using a dated, descriptive name (e.g. <project>_<subject>_架构图_YYYYMMDD.svg). Always visually verify the exported PNG before presenting.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

