Skip to content
Get packs

FigureSpec Architecture Diagrams

Turns structured JSON into editable, deterministic SVG architecture and workflow diagrams.

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
2 files

What it does

When installed, the agent can turn a description of a system, pipeline or workflow into a structured JSON spec and then render it locally into an SVG vector diagram. Because the output is plain-text SVG, the same spec always produces the same figure and the result can be edited by hand afterwards. It is aimed at formal diagrams for academic papers and technical documents, not charts or artistic illustrations.

Great for

  • Researchers who need clean, editable architecture or pipeline figures for a paper
  • Developers documenting system layers, workflows or audit cascades as vector graphics
  • Anyone who wants reproducible diagrams that can be version-controlled as text

Skip it if

  • Data plots such as bar, line or scatter charts, which the skill itself says should use a different skill (reason: wrong tool for the job)
  • Quick flowcharts where a lighter syntax like Mermaid is enough (reason: this skill needs a full JSON spec and more setup)
  • Natural or illustrative images (reason: it only draws structured diagrams with boxes and arrows)

Try asking

  • “Draw a layered architecture diagram of our data ingestion, training and serving components as an SVG”
  • “Make a left-to-right workflow figure showing how the pipeline steps produce and verify results”
  • “Create a hub-and-spoke topology diagram from this description for my paper's system overview”
Step-by-step guide: how to use FigureSpec Architecture DiagramsInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 2 original files, the MIT licence and an attribution note.

    figure-spec.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip figure-spec.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip figure-spec.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Describe the diagram's entities, connections and groupings clearly so the agent can build a good JSON spec on the first try.
  • Run the validate command on the spec before rendering to catch schema errors early.
  • Open the generated SVG in a vector editor or text editor to fine-tune labels and positions.

What's inside

  • scripts/figure_renderer.py
  • SKILL.md

Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.