Who it is forA developer who has just joined or taken over a project and needs its overall structure quickly.
The problemReading a large codebase file by file takes days and hides how the main modules depend on each other.
The outcomeAn interactive architecture view in index.html that you can search, zoom, and open offline, with node and edge source references to check.
What you need first
- Node.js 22 or later
- A plugin-capable client such as Claude Code, Codex, Cursor or Qoder, with model access configured
- A local checkout of the application project you want to diagram
Steps
- Run
npx skills add supermax92/qgraphflowto install the q-flow skill. - Open your application project in your client and start a new session.
- In Claude Code, enter
/q-flow Help me draw this project. I want to understand it as quickly as possible. - Open the generated index.html in your browser and use search, zoom and pan to explore responsibilities and upstream or downstream relationships.
- Inspect a few nodes and edges to confirm their source files, lines and symbols before relying on the diagram.
Prompt for your agent
Use the QGraphFlow q-flow skill on the project at <project-path>. Generate an architecture diagram showing the main modules, their dependencies and system boundaries, and write it under docs/qgraphflow/. Verify that the output folder contains index.html and graph.json, and report any relationships the diagram marks as uncertain.
Caveat. The skill runs inside a plugin-capable client with model access, so you need one configured before any diagram can be generated.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for supermax92/qgraphflow and "Map an unfamiliar codebase in one offline diagram"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer