blueprint-mcp-diagram-generation — Claude Code skill
blueprint-mcp-diagram-generation turns a written description of a system into architecture diagrams, flowcharts or data-flow pictures, using Nano Banana Pro behind an Arcade MCP server.
no license 189 installs ★ 9 updated 2026-10-09
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a claude-codeWhat blueprint-mcp-diagram-generation does
The skill hands the agent a routine for the Blueprint MCP server: open a job with a detailed prompt through start_diagram_job, poll check_job_status until it reports Complete, then fetch the base64 PNG with download_diagram and decode it into the project folder. Jobs run in the background, moving from pending to processing and then to a final state.
Nothing gets drawn until the server exists. You need an Arcade account and a Google AI Studio key stored as the GOOGLE_API_KEY secret. Deploy the blueprint-mcp repo with arcade-mcp deploy, add the server to a gateway in the Arcade dashboard, and paste the gateway URL into the mcpServers block of your editor config. Skip that last step and the agent has no tool to call.
Prompts work best layer by layer: what sits on each tier, which arrows carry labels, which style and aspect ratio you want. Vague descriptions come back as vague pictures.
What it helps with
- map a microservice fleet into one labeled diagram
- draw an OAuth login sequence including the failure branch
- show an ETL pipeline from sources to warehouse
- save the finished PNG next to the code
- explain an unfamiliar repo through a single picture
Example prompts for the agent
- Draw the architecture of this service: gateway, auth, orders and payments, with labeled arrows
- Make a sequence diagram for the OAuth 2.0 login flow, including the token validation error path
- Diagram the data flow of the ETL job and save it to
architecture_diagram.png
How to install blueprint-mcp-diagram-generation
1Skills CLI (Vercel)
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a claude-code2Manually: put the files in place
~/.claude/skills/blueprint-mcp-diagram-generation/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/blueprint-mcp-diagram-generation/
1Skills CLI (Vercel)
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a codex2Manually: put the files in place
.agents/skills/blueprint-mcp-diagram-generation/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a cursor2Manually: put the files in place
.cursor/skills/blueprint-mcp-diagram-generation/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install reason-machines/mcp-skills skills/blueprint-mcp-diagram-generation2Manually: put the files in place
.github/skills/blueprint-mcp-diagram-generation/SKILL.md1Skills CLI (Vercel)
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a gemini-cli2Platform CLI
gemini skills install https://github.com/reason-machines/mcp-skills --path skills/blueprint-mcp-diagram-generationManually: put the files in place
.devin/skills/blueprint-mcp-diagram-generation/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about blueprint-mcp-diagram-generation
How do I install blueprint-mcp-diagram-generation in Claude Code?
Run npx skills add reason-machines/mcp-skills --skill blueprint-mcp-diagram-generation -a claude-code. The skill lands at ~/.claude/skills/blueprint-mcp-diagram-generation/SKILL.md.
What does blueprint-mcp-diagram-generation do?
It gives the agent a procedure for producing architecture diagrams, flowcharts and data-flow pictures through the Blueprint MCP server, which runs Nano Banana Pro. The agent starts the job, polls it and downloads the PNG.
What do I need before diagrams actually render?
An Arcade account and a Google AI Studio API key saved with arcade-mcp secret set GOOGLE_API_KEY="your_api_key_here". The deployed server also has to sit in a gateway, with that gateway URL in your MCP config.