markdown-viewer/skills: 14 Agent Skills for Diagrams Inside Markdown
Opinionated skills for AI coding agents to create stunning diagrams and visualizations directly in Markdown. These skills extend agent capabilities across diagram generation, data visualization, and technical documentation.
At a glance
- What is it?
- The repository packages fourteen opinionated skills that let coding agents emit charts, UML, cloud and network diagrams as Markdown code fences. It is a content library rather than a renderer, and the README documents installation far better than it documents what happens when an engine is missing.
- Who is it for?
- Adopt markdown-viewer/skills if your team already writes Markdown in Claude Code, Codex, Cursor or VS Code and wants diagrams to come out of the same agent session; skip it if you need a standalone renderer, a documented support policy, or a licence you can read before shipping.
- Can I use it commercially?
- Yes, with credit. CC-BY-4.0 allows commercial use as long as you credit the authors and indicate what you changed. It is written for creative content, so check how it applies to any code.
- Is it still maintained?
- Yes. The repository last received commits 41 days ago.
- What is it written in?
- GitHub does not report a main language for this repository.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- markdown-viewer/skills3,404 ★ · +57
| Repository | 2026-09-28 | 2026-11-15 |
|---|---|---|
| markdown-viewer/skills | 3.3k | 3.4k |
57 Stars in the last 48 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
What markdown-viewer/skills actually ships
This is a collection of prompt-and-template packages for AI coding agents, not a diagramming application. The README describes fourteen skills covering five rendering engines, spanning software modeling, enterprise architecture, data analytics and editorial content cards. Each skill lives in its own top-level directory (archimate/, architecture/, bpmn/, canvas/, cloud/, data-analytics/, iot/, infocard/, infographic/, mindmap/, network/, security/, uml/, vega/), and each carries a SKILL.md that the agent reads. The format is the Agent Skills convention, linked from the README, so the unit of distribution is a folder with instructions inside it rather than a library you import.
The practical audience is people who already have an agent writing Markdown for them and want the output to include something other than prose and tables. A backend engineer documenting a service boundary, a consultant producing a cloud architecture sketch, or a technical writer assembling a KPI card all fit. Someone who wants to draw a diagram by hand in a GUI does not.
How the skills reach the model: folders, fences and engines
The mechanism is indirect, and that is the most important thing to understand before installing anything. The repository does not render diagrams. It supplies instructions that teach an agent which syntax to emit for a given request, and the resulting diagram is a fenced code block in the Markdown file.
The skills split into three families. Standalone skills emit their own fence language: vega for Vega-Lite and Vega charts, infographic for seventy-plus YAML-based templates, canvas for JSON Canvas node diagrams. HTML/CSS embedded skills (architecture, infocard) generate HTML and CSS directly inside the Markdown with no code fence at all. The PlantUML family (uml, cloud, network, security, archimate, bpmn, data-analytics, iot, mindmap) all emit ```plantuml or ```puml blocks and lean on domain-specific mxgraph stencil icons. The UML skill alone is documented as covering fourteen diagram types with more than 9500 stencil icons.
The selection guide in the README is the real interface. It maps a use case to a skill: a sequence diagram goes to uml, a KPI dashboard to infographic, a threat model to security, a free-position mind map to canvas while a hierarchical one goes to mindmap. That mapping is opinionated on purpose. The trade-off is that the agent has to pick correctly, and nothing in the repository enforces the choice.
Installing the skills and producing a first diagram
The README's recommended path is a single npx command that it says works across multiple agents, including Claude Code, Codex and Cursor.
npx skills add markdown-viewer/skillsAfter it runs, the skills are available to the agent. If you prefer to place one skill by hand for Claude Code, the README gives a copy command into the user-level skills directory.
cp -r skills/<skill-name> ~/.claude/skills/For GitHub Copilot or VS Code, the README states that skills are detected automatically when placed in a .github/skills/ directory. For claude.ai, the instruction is to add the skills to project knowledge or paste the contents of SKILL.md into the conversation.
Once a skill is in place, the first real use is a prompt, not a command. Ask the agent for the diagram you want and it should emit the fence that skill documents. A Vega-Lite chart, for example, arrives as a vega-lite block that your Markdown viewer has to render. The README does not ship a renderer, so the thing to check immediately is whether your viewer understands the fence languages the skills produce.
Where the repository is thin: no renderer, no licence, no releases
The README is a catalogue, and the gaps matter more than the catalogue does. It documents installation for four agent environments and then stops. There is no section on what renders the output, no troubleshooting for a viewer that shows a raw PlantUML block, and no statement about which Markdown viewers were used to check the templates. If your viewer does not support the fence language, the skill has still worked and you still get nothing to look at.
The repository also has no licence file listed among its top-level entries, and the licence field is unknown. For a library of templates and stencil conventions that people may embed in client deliverables or commercial documentation, that is a real adoption blocker, and it is not something a README can fix by asserting permissiveness. The repository has no published GitHub releases either, so there is no versioned artifact to pin; consumers take whatever the default branch holds.
Maintenance is the other open question. The last push was on 2026-05-26. The repository is not archived, but the README does not describe a support policy, a compatibility matrix for agent versions, or a deprecation process for skills that stop working when an engine changes.
PlantUML skills versus a diagram-as-code toolchain
The closest alternative for the PlantUML-based skills is a conventional diagram-as-code setup such as Mermaid or PlantUML driven by a CLI and a CI job. The difference is where the intelligence sits. With a CLI toolchain, a human writes the diagram source and the tool renders it deterministically; the syntax is the interface and the renderer is the contract. With markdown-viewer/skills, the agent writes the source from a natural-language request, and the skill folder is what steers it toward stencils, layouts and conventions.
That shift buys speed on first drafts and costs predictability. A CI pipeline that renders Mermaid will fail loudly on bad syntax; an agent that emits a PlantUML block with an icon name the stencil set does not contain produces something that looks plausible and renders wrong, and the repository's skill files are the only place that constraint is expressed. The skills also assume a Markdown-centric workflow. If your diagrams live in a dedicated tool and get exported as images, the fence-based output is a mismatch rather than a feature.
Licence and upgrade cost before you commit
The licence is not stated in the repository's top-level entries, and no licence file appears among them. That means you cannot confirm the terms under which the skill files, the infographic templates or the mxgraph stencil conventions may be redistributed or embedded in a product. Treat that as unresolved rather than permissive. If you plan to ship generated diagrams inside a commercial deliverable, resolve the licence question first; this is not legal advice, and a lawyer should read whatever terms exist.
Upgrade cost is shaped by the delivery method. The recommended npx install pulls from the repository, so there is no version to pin and no changelog to read between updates. A skill that changes its template set or its fence convention can alter agent output without any local diff. The manual copy into ~/.claude/skills/ or .github/skills/ gives you a snapshot you control, at the cost of re-copying by hand when you want a newer one. If you need reproducibility, the manual path is the one that gives it to you.
Editorial conclusion
Adopt markdown-viewer/skills if your team already writes Markdown in Claude Code, Codex, Cursor or VS Code and wants diagrams to come out of the same agent session; skip it if you need a standalone renderer, a documented support policy, or a licence you can read before shipping. Before committing, install one skill with npx skills add markdown-viewer/skills, generate a single diagram with the code fence that skill documents, and confirm your Markdown viewer actually renders PlantUML and Vega blocks, because the repository does not ship a fallback for viewers that do not.
Frequently asked questions
How do I install markdown-viewer/skills in Claude Code?
The README recommends the quick install with npx skills add markdown-viewer/skills, which it says works across multiple agents including Claude Code. Alternatively you can copy a single skill folder into ~/.claude/skills/ with cp -r skills/<skill-name> ~/.claude/skills/.
How do I install markdown-viewer/skills in Codex?
The README states that the quick install method, npx skills add markdown-viewer/skills, works with multiple AI coding agents including Codex. No Codex-specific manual path is documented.
How do I use markdown-viewer/skills in Claude Code?
Install the skills, then ask the agent for the diagram you want; the skill instructs it to emit the matching code fence, such as a plantuml or vega-lite block. The README's skill selection guide maps use cases to skills, for example uml for sequence diagrams and infographic for KPI dashboards.
Can I install markdown-viewer/skills from GitHub into Claude?
Yes. The recommended command, npx skills add markdown-viewer/skills, installs directly from the repository, and the README also documents copying a skill folder into ~/.claude/skills/ by hand.
How do I install markdown-viewer/skills in ChatGPT?
The README does not document a ChatGPT installation path. It lists quick install for Claude Code, Codex and Cursor, manual copies for Claude Code and claude.ai, and automatic detection for GitHub Copilot and VS Code.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/markdown-viewer-skills)Star history chart for your README
[](https://hysenlabs.com/projects/markdown-viewer-skills)Was this useful?