← Back to learning path
Case · Published

Codex x Draw.io MCP: let AI draw architecture diagrams

Use Codex with Draw.io MCP to turn system relationships into editable architecture or flow diagrams while keeping installation and permission boundaries explicit.

Source: recipes/02-drawio-mcp.md实战案例 (original language)
Chapter progress 0%Learning path progress 0% · 0/46 chapters completedContinue learning path →

Tip · Last checked

Official sources were checked on 2026-09-04. The MCP mechanism follows the Codex MCP documentation. This case uses the community repository jgraph/drawio-mcp. See the draw.io website for the editor. Follow the third-party repository for current installation commands, authorization, and compatibility.

Codex x Draw.io MCP: let AI draw architecture diagrams

Diagrams make system relationships easier to inspect than a long block of prose. Draw.io can represent flow charts, architecture diagrams, mind maps, system designs, business processes, and technical roadmaps.

This case shows how to ask Codex to install and use the Draw.io MCP tool to create an editable diagram. The repository is jgraph/drawio-mcp.

drawio-mcp-github-repository

1. Install

You can add the server in settings by filling in its configuration fields.

drawio-mcp-add-server-settings

An easier approach is to give Codex the GitHub URL and ask it to complete the installation. If it needs a clearer instruction, say: "Install it and make sure I can use it directly."

drawio-mcp-install-prompt

After installation, review the command result and confirm the server before using it.

drawio-mcp-install-command-result

2. Use it

Give Codex a bounded task, such as turning the key ideas of a book into a Chinese information diagram.

drawio-mcp-permission-model-prompt

Specify the content, theme, and visual style you need. Explicitly mention Draw.io MCP so the agent selects the intended tool.

drawio-mcp-permission-model-source-links

Codex can summarize the run and open the diagram in a browser after the tool completes.

drawio-mcp-read-write-own-summary

drawio-mcp-read-write-own-diagram

References

The operation flow and demonstration scenario were also informed by this creator's video:

The screenshots are from the author's own operation. The text has been reorganized and rewritten; contact the site owner for removal requests.

Further reading

KnowledgeModel Serving in Production: GPU Inference, Queues, and ScalingKnowledgeProduction Prompt Engineering: From Instructions to Measurable Contracts
Product Hub Demo · V0.3