How do I use MCP Demo Server?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@MCP Demo Server list available tools" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

MCP Demo Server

by jlnovais

Overview Schema Related Servers Score Discussions

TypeScript

Remote

MCP Demo

A monorepo for experimenting with the Model Context Protocol (MCP). It exposes the MindShaker Wallet API as MCP tools through a NestJS server, plus Claude clients (CLI and web) that connect to the server and call those tools. The server also includes RAG search over a local knowledge base.

Project structure

MCP-demo/
├── apps/
│   ├── mcp-server/    # NestJS app — MCP server (Wallet API + RAG + utilities)
│   └── mcp-client/    # Claude CLI + React/Express web chat client
├── libs/
│   └── api-common/    # Shared config utilities
├── docs/              # Guides (setup, RAG, deploy, Claude Desktop, …)
├── package.json
└── README.md

Related MCP server: MCP Redirect Server

Prerequisites

Node.js 18+ (22 recommended)
npm 9+
A running MindShaker Wallet API (for Wallet tool calls)
An Anthropic API key (for either client)
For RAG: a Voyage AI API key and a vector store (Postgres + pgvector by default, or LanceDB)

Getting started

For the full walkthrough (env vars, CLI vs web, troubleshooting), see docs/getting-started.md.

Quick path from the repository root:

npm install
cp apps/mcp-server/.env.template apps/mcp-server/.env
cp apps/mcp-client/.env.template apps/mcp-client/.env

mcp-server (`apps/mcp-server/.env`)

Variable	Description
`PORT`	MCP server port (default `4000`)
`MCP_SERVER_API_KEY`	Key required to call the MCP endpoint (`x-api-key` or `Authorization: Bearer`)
`WALLET_API_BASE_URL`	Upstream Wallet API base URL
`WALLET_API_KEY`	Wallet API key sent as `x-wallet-api-key` when tools call the upstream API
`VOYAGE_API_KEY`	Voyage AI key for `search_knowledge_base`
`VOYAGE_EMBED_MODEL`	Embedding model (default `voyage-4`)
`VECTOR_STORE`	`postgres` (default) or `lancedb`
`POSTGRES_*`	Postgres connection when `VECTOR_STORE=postgres`
`LANCEDB_*`	LanceDB path/table when `VECTOR_STORE=lancedb`

mcp-client (`apps/mcp-client/.env`)

Variable	Description
`ANTHROPIC_API_KEY`	Required. Anthropic API key
`MCP_SERVER_URL`	MCP server endpoint (default `http://127.0.0.1:4000/mcp/v1`)
`MCP_SERVER_API_KEY`	Must match the server's `MCP_SERVER_API_KEY`
`CLAUDE_MODEL`	Claude model (default `claude-sonnet-5`)
`CLAUDE_MAX_TOKENS`	Max output tokens (default `4096`)
`CLAUDE_THINKING_BUDGET`	Extended-thinking budget; set `0` to disable
`CLIENT_WEB_PORT`	Web UI port (default `3001`)
`CLIENT_WEB_USERNAME` / `CLIENT_WEB_PASSWORD`	Required for web client — HTTP Basic Auth

Start the MCP server (terminal 1):

npm run start:server:dev

When ready, the server prints:

MCP server is running on port 4000
MCP endpoint: POST http://localhost:4000/mcp/v1

Then start a client (terminal 2):

npm run start:client          # CLI chat
npm run start:client:web      # Browser UI at http://localhost:3001

Scripts

Run these from the repository root:

Script	Description
`npm run start:server`	Start the NestJS MCP server
`npm run start:server:dev`	Start the server with hot reload
`npm run start:client`	Interactive Claude CLI client
`npm run start:client:web`	Web chat UI (dev mode with Vite HMR)
`npm run start:client:web:prod`	Web UI serving a pre-built static bundle
`npm run build:client:web`	Build the React UI
`npm run build`	Build all workspaces
`npm run lint`	Lint TypeScript across apps and libs
`npm run format`	Format TypeScript with Prettier
`npm run test`	Run Jest tests

You can also run scripts inside a single app with the -w flag:

npm run start:dev -w @mcp-demo/mcp-server
npm run ingest:knowledge -w @mcp-demo/mcp-server
npm run build -w @mcp-demo/mcp-client

Apps

`@mcp-demo/mcp-server`

A NestJS application that exposes the MindShaker Wallet API as MCP tools over HTTP, plus RAG and date utilities.

Port: 4000 (override with PORT)
MCP endpoint: POST /mcp/v1 (requires x-api-key or Authorization: Bearer)
Source: apps/mcp-server/src/

MCP tools

Category	Tool	Backend
Payments	`create_payment`	`POST /api/payments`
	`list_payments`	`GET /api/payments`
	`get_payment`	`GET /api/payments/{id}`
	`cancel_payment`	`DELETE /api/payments/{id}/cancel`
Wallets	`update_wallet`	`PUT /api/wallet`
	`get_wallet`	`GET /api/wallet/{userId}/{merchantId}`
	`transfer_credits`	`PUT /api/wallet/transfer`
	`reset_wallets`	`PUT /api/wallet/reset`
	`list_wallet_logs`	`GET /api/wallet/logs`
	`get_wallet_log`	`GET /api/wallet/logs/{id}`
Exchange rate	`upsert_exchange_rate`	`POST /api/exchange-rate`
	`get_exchange_rate`	`GET /api/exchange-rate/{merchantId}`
Knowledge base	`search_knowledge_base`	Semantic search over local docs (RAG)
Utilities	`get_current_datetime`	Server UTC datetime
	`add_duration_to_datetime`	Shift an ISO datetime by a duration

Tool registration lives under apps/mcp-server/src/mcp/tools/. Each Wallet category has a matching service in apps/mcp-server/src/mcp/api/.

The search_knowledge_base tool adds Retrieval-Augmented Generation (RAG) over your own documents using LlamaIndex + Voyage AI embeddings + a vector store (Postgres + pgvector by default, or LanceDB). Configure VOYAGE_API_KEY and VECTOR_STORE, add documents under apps/mcp-server/knowledge/, then build the index:

npm run ingest:knowledge -w @mcp-demo/mcp-server

See docs/rag-knowledge-base.md for full setup.

`@mcp-demo/mcp-client`

Claude clients that talk to Anthropic and expose the MCP server's tools to the model. Both connect over streamable HTTP, discover tools, and let Claude call them while you chat.

CLI: npm run start:client — terminal chat (apps/mcp-client/src/)
Web: npm run start:client:web — React UI + Express API on port 3001 (apps/mcp-client/web/, apps/mcp-client/src/server.ts)
Transport: Streamable HTTP to the MCP server (MCP_SERVER_URL)

Type a message and press Enter (CLI), or open http://localhost:3001 (web; Basic Auth from .env). Type exit or press Ctrl+C to quit the CLI.

Docs

Guide	Topic
getting-started.md	Install, env, start server/clients
testing-mcp-endpoint.md	curl examples for the MCP HTTP API
claude-desktop.md	Claude Desktop integration
rag-knowledge-base.md	RAG setup and ingestion
rag-search-and-chunking.md	Search and chunking details
postgres-knowledge-schema.md	Postgres + pgvector schema
server-knowledge-ingestion.md	Knowledge ingestion on the server
deploy/DEPLOY.md	Deploy with PM2 on Ubuntu

Workspaces

This project uses npm workspaces. Dependencies are hoisted to the root node_modules/, and each app under apps/ has its own package.json.

License

ISC

This server cannot be installed

license - permissive license

quality - not tested

maintenance

How are these scores calculated?

Maintenance

–Maintainers

–Response time

–Release cycle

–Releases (12mo)

Commit activity

Resources

Need Help?

Related Servers

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

NestJS MCP Server
Developer Tools Code Execution
RenzoReccio
A
license
-
quality
D
maintenance
A demonstration implementation of Model Context Protocol (MCP) using NestJS framework, allowing developers to build MCP-compatible applications with playground testing capabilities.
Last updated 2025-07-29
8
MIT
MCP Redirect Server
Developer Tools Autonomous Agents
marcos2872
F
license
-
quality
D
maintenance
A Model Context Protocol server built with NestJS that provides OAuth 2.1 authentication with GitHub and exposes MCP tools through Server-Sent Events transport. Enables secure, real-time communication with JWT-based protection and dependency injection.
Last updated 2025-10-27
Learn MCP Server
Developer Tools
saravmani-kmu
F
license
-
quality
D
maintenance
A Model Context Protocol (MCP) server designed for learning and experimentation. It provides a foundational setup for developers to build, run, and debug MCP server implementations using Node.js.
Last updated 2026-03-06
mcp-nestjs
Developer Tools App Automation
zahidhasanaunto
A
license
-
quality
C
maintenance
A NestJS module for building Model Context Protocol (MCP) servers using decorators to expose services as tools, resources, and prompts. It features auto-discovery, a built-in playground UI, and support for multiple transports including SSE and Stdio.
Last updated 2026-04-08
12
MIT

View all related MCP servers

Related MCP Connectors

mcp
MCP (Model Context Protocol) server for Appwrite
hithereiamaliff-mcp-nextcloud
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
console-mcp-server
The official MCP Server from Mia-Platform to interact with Mia-Platform Console

View all MCP Connectors

Latest Blog Posts

Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)
By Om-Shree-0709 on July 25, 2026.
mcp
Agent Identity
OAuth 2.1
Your AI Chatbot Just Exposed Your CEO's Salary to an Intern
By Om-Shree-0709 on July 2, 2026.
Agent Identity
MCP Security
OAuth Delegation
Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)
By Om-Shree-0709 on June 30, 2026.
Agentic Ai
Prompt Injection
WebAssembly

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/jlnovais/MCP-demo'

If you have feedback or need assistance with the MCP directory API, please join our Discord server