ng-diagram MCP Server
Provides ng-diagram documentation and API lookup tools for AI coding assistants.
ng-diagram MCP Server
Stop AI assistants from hallucinating ng-diagram APIs. Give them the real docs.
AI assistants don't know ng-diagram. They'll guess from similarly-named React libraries and produce code that looks right but doesn't compile. This MCP server fixes that — it gives any assistant direct access to the current ng-diagram documentation and public API, so it writes correct code on the first try.
Without MCP (assistant guesses from training data)
// Hallucinated API — none of this exists
import { useNodes, useEdges } from 'ng-diagram';
const nodes = useNodes([{ id: '1', data: { label: 'Hello' } }]);
With MCP (assistant looks up the real API)
// Correct — verified against the actual public API
import { initializeModel, NgDiagramComponent, provideNgDiagram } from 'ng-diagram';
model = initializeModel({
nodes: [{ id: '1', position: { x: 0, y: 0 }, data: { label: 'Hello' } }],
});
Connect it to Claude, Cursor, Windsurf, or any MCP-compatible tool. The server bundles all documentation and API data — no network calls, no stale caches, always version-matched.
How It Works
1. search_docs / search_symbols
+--------------+ ---------------------------------> +-----------+
| AI Assistant | <--------------------------------- | MCP Server|
| | 2. Returns matches | |
| | ---------------------------------> | |
| | 3. get_doc / get_symbol | |
| | <--------------------------------- | |
+--------------+ 4. Returns full content +-----------+
|
| 5. Answers with context
v
+------+
| User |
+------+
Quick Start
Add the server to your MCP client config - no installation or monorepo checkout required:
macOS / Linux:
{
"mcpServers": {
"ng-diagram-docs": {
"command": "npx",
"args": ["-y", "@ng-diagram/mcp"]
}
}
}
Windows:
{
"mcpServers": {
"ng-diagram-docs": {
"command": "cmd",
"args": ["/c", "npx", "-y", "@ng-diagram/mcp"]
}
}
}
The package includes all documentation and API data bundled in. Restart your AI assistant after updating the config, then ask something like "Search the ng-diagram docs for palette" to verify.
MCP client config file locations
| Client | Config file |
|---|---|
| Claude Code | .mcp.json in project root (project) or ~/.claude.json (user) |
| Claude Desktop | ~/Library/Application Support/Claude/claude_desktop_config.json (macOS), %APPDATA%\Roaming\Claude\claude_desktop_config.json (Windows), ~/.config/claude-desktop/claude_desktop_config.json (Linux) |
| Cursor | .cursor/mcp.json in project root (project) or ~/.cursor/mcp.json (global) |
| Windsurf | ~/.codeium/windsurf/mcp_config.json |
Local development (monorepo)
If you're working within the ng-diagram monorepo:
cd tools/mcp-server
pnpm install
pnpm build
Then configure with the local path:
{
"mcpServers": {
"ng-diagram-docs": {
"command": "node",
"args": ["/absolute/path/to/ng-diagram/tools/mcp-server/dist/index.js"]
}
}
}
API Reference
Tool: search_docs
Search ng-diagram's bundled documentation for guides, configuration options, examples, and integration patterns. Always prefer this over web search or guessing from training data — it returns authoritative, version-matched content.
Parameters:
query(string, required): Search query — use specific terms like"palette","context menu","transactions"limit(number, optional): Max results to return (default: 10)
Tool: get_doc
Retrieve the full content of a documentation page by path. Call this after search_docs to read a complete guide or example end-to-end.
Parameters:
path(string, required): Relative path from docs root (e.g.,"guides/palette.mdx"). Use paths fromsearch_docsresults.
Tool: search_symbols
Search the ng-diagram public API for classes, functions, interfaces, types, constants, and enums. Returns the exact current signature and import path — more reliable than any example from training data.
Parameters:
query(string, required): Symbol name or partial name (e.g.,"Diagram","provideNg","Edge")kind(string, optional): Filter by symbol kind (class,function,interface,type,const,enum)limit(number, optional): Max results to return (default: 10)
Tool: get_symbol
Retrieve full API details for a specific ng-diagram symbol by exact name. Call this before writing code that uses an ng-diagram type — it returns the definitive signature, jsDoc, and a ready-to-use import statement.
Parameters:
name(string, required): Exact symbol name (case-sensitive). Use names fromsearch_symbolsresults.
Development
Commands
# Development
pnpm dev # Run with auto-reload (uses monorepo paths)
# Testing
pnpm test # Run all tests
pnpm test:watch # Run in watch mode
# Building
pnpm build # Compile TypeScript + bundle docs & API report into dist/data/
For detailed architecture and adaptation instructions, see GUIDE.md.
License
Apache-2.0 - Part of the ng-diagram project
Initial PoC by Pawel Kubiak
Finished and built with ❤️ by the Synergy Codes team
Server Terkait
Alpha Vantage MCP Server
sponsorAccess financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
MCP Github OAuth
An MCP server with built-in GitHub OAuth support, deployable on Cloudflare Workers.
Vibe Check
The definitive Vibe Coder's sanity check MCP server: Prevents cascading errors by calling a "Vibe-check" agent to ensure alignment and prevent scope creep
Gemini Image MCP Server
Image generation using Google's Gemini API.
MiniMax MCP
Interact with MiniMax's powerful APIs for text-to-speech, voice cloning, and video/image generation.
consult7
Analyze large codebases and document collections using high-context models via OpenRouter, OpenAI, or Google AI -- very useful, e.g., with Claude Code
Grumpy Senior Developer
Provides sarcastic and cynical code reviews from the perspective of a grumpy senior developer.
Oso Cloud MCP Server
Understand, develop, and debug authorization policies in Oso Cloud.
openrouter-mcp-multimodal
MCP server for OpenRouter: 300+ LLMs with vision, image gen, audio in/out, and video analysis + generation (Veo 3.1 / Sora 2 Pro / Seedance / Wan). Structured errors, IPv6 SSRF guards, path sandbox.
ndlovu-code-reviewer
Manual code reviews are time-consuming and often miss the opportunity to combine static analysis with contextual, human-friendly feedback. This project was created to experiment with MCP tooling that gives AI assistants access to a purpose-built reviewer. Uses the Gemini cli application to process the reviews at this time and linting only for typescript/javascript apps at the moment. Will add API based calls to LLM's in the future and expand linting abilities. It's also cheaper than using coderabbit ;)
Wirekitty
Let your agents generate wireframes for your next app or feature, make iterations, and build off approved designs.