Skip to main content

Draw diagrams in Claude and ChatGPT with the ZenUML render server

The ZenUML render server lets an AI assistant draw a diagram for you right in the conversation. Ask Claude, ChatGPT or another client that supports the Model Context Protocol for a sequence diagram of a login flow or a flowchart of a deploy. The assistant writes the diagram source, and ZenUML draws it.

It needs no account and no sign-in, and it saves nothing: the diagram exists only in your conversation. To save diagrams to Confluence pages, use the ZenUML MCP server for Confluence instead.

To keep the diagram source off our servers entirely, run the local ZenUML MCP server on your own computer instead.

Server URL: https://zenuml.com/mcp

Existing connections to https://conf-lite.zenuml.com/mcp or https://conf-lite.zenuml.com/mcp/render continue to work.

Supported diagrams

TypeUse it forSource
SequenceCalls and messages between services or peopleZenUML
MermaidFlowcharts, state, class, entity-relationship, Gantt and other Mermaid diagramsMermaid
PlantUMLAny PlantUML diagramPlantUML, between @startuml and @enduml
GraphDrawings you already have as draw.io XMLdraw.io / diagrams.net XML
OpenAPIREST API documentationOpenAPI or Swagger, in YAML or JSON
AsyncAPIEvent-driven API documentationAsyncAPI 2.x or 3.x, in YAML or JSON
MarkdownNotes and documents, with Mermaid blocks drawn as diagramsMarkdown; the assistant must say it is Markdown

For every other type, ZenUML detects the language from the source when the assistant doesn't name it.

Set it up

In Claude Code

Install the ZenUML plugin, which adds the render server and a skill that tells Claude how to use it:

/plugin install zenuml --marketplace ZenUml/zenuml-plugin

Or add only the server:

claude mcp add --transport http zenuml-render https://zenuml.com/mcp

In other clients

Add a remote MCP server with the URL https://zenuml.com/mcp. It uses streamable HTTP and needs no authentication. Clients that support MCP Apps, such as Claude, show the drawn diagram inline; other clients receive the result as text.

Use it

Ask in plain language, for example:

  • "Draw a sequence diagram of an OAuth login with PKCE."
  • "Turn these deploy steps into a flowchart: build, test, deploy to staging, deploy to production."
  • "Draw this OpenAPI spec as API documentation."
  • "Turn this AsyncAPI spec into documentation I can read."

If the source doesn't parse, ZenUML returns each error with its line and column, and the assistant can fix the source and draw it again. Ask for the source if you want to keep it, or paste it into a ZenUML macro in Confluence.

What the assistant can do

ToolWhat it does
render_diagramDraws a diagram from its source and reports whether the source parsed. Read-only; nothing is saved.

The server also offers a short writing guide for each language, as the resources zenuml://dsl-guide, mermaid://dsl-guide, plantuml://dsl-guide and openapi://dsl-guide.

Limits

  • Rate limit: 30 diagrams per minute from one network address. Beyond that, the assistant is told how many seconds to wait.
  • Source size: up to 500,000 characters per diagram.

Privacy

  • Diagram source: ZenUML checks it and sends it back to your assistant to be drawn; it doesn't store it.
  • PlantUML: drawn in the view by a built-in engine. No PlantUML server receives the source.
  • Rate limiting: counts calls against a one-way hash of your network address, never the address itself.
  • Usage analytics: records no account or user identity.

Your assistant sends the conversation, including the diagram source, to its own AI model. Before you draw diagrams of confidential systems, check that this is allowed where you work.

For details, see the privacy policy.