NextFlows Academy MCP Starter
A starter MCP server with a working 'greet' tool and stub examples for notes/FAQ search, built for the NextFlows Academy cohort program to extend with custom tools.
README
MCPRepo — NextFlows Academy Starter
Part of NextFlows Academy — the free cohort program Building an MCP for an AI Engine.
Clone this repo to build your Model Context Protocol (MCP) server in TypeScript. By Demo Day you will ship a public GitHub repo with real tools, Zod validation, docs, and a live demo — the same path used in the free NextFlows Academy cohort.
Program hub: nextflows.ai/academy
Full program page (in this repo): docs/PROGRAM.md
Apply: Cohort application
About NextFlows Academy
NextFlows Academy runs structured, cohort-based programs with live sessions, mentor support, and a real project you ship by the end.
This repository belongs to:
| Program | Building an MCP for an AI Engine |
| Audience | 4th & 5th year CS / CE students |
| Duration | 6 weeks |
| Format | Cohort + project |
| Price | Free |
| Level | Intermediate |
| Outcome | Shipped MCP server on GitHub |
| Schedule | Wed & Sat online 1:30–3:30 PM + Monday on-site workshop days |
You go from “what’s an MCP?” to a working MCP server connected to an AI engine (for example Claude), fully documented, and live on GitHub.
See docs/PROGRAM.md for outcomes, weekly plan, starter projects, and who it’s for.
What you get
| Path | Purpose |
|---|---|
src/index.ts |
MCP server + stdio transport |
src/tools/ |
One register helper per tool |
src/schemas/ |
Zod input contracts (with .describe(...)) |
examples/ |
Sample JSON args for Inspector |
docs/PROGRAM.md |
Full NextFlows Academy program page |
docs/WEEK-2.md |
Full Week 2 step-by-step plan |
docs/CURRICULUM.md |
6-week overview |
docs/project-choice.md |
Week 2 project choice template |
docs/design.md |
Week 2 design doc template |
Week 1 is already wired: a working greet tool so you can open Inspector on day one.
Week 2 examples included: stub tools for Notes & FAQ Search (search_notes, list_notes, add_note). Enable them when you pick that starter (or copy the pattern for your own idea).
Prerequisites
- Node.js 20+ (
node -v) - npm (
npm -v) - Git + a GitHub account
- Cursor or VS Code
Quick start
git clone <YOUR_FORK_OR_ORG_URL>/MCPRepo.git
cd MCPRepo
npm install
npm run inspect
In the Inspector browser tab:
- Click Connect
- Open Tools
- Select
greetand putAlexin the name field (seeexamples/greet.jsonfor the full args shape) - Try invalid input (empty name) and confirm Zod rejects it
To run the server alone (waits on stdin):
npm run dev
Important: log only with
console.error. Never useconsole.log— stdout is reserved for the MCP protocol.
Week 2
Week 2 is design-first. Follow docs/WEEK-2.md.
Useful scripts:
| Script | What it does |
|---|---|
npm run dev |
Start the MCP server on stdio (stays alive; stop with Ctrl+C) |
npm start |
Same as dev |
npm run inspect |
Open MCP Inspector against this server |
Stack
- TypeScript via
tsx(no build step early on) - Official MCP TypeScript SDK (
@modelcontextprotocol/server) - Zod for tool
inputSchema - MCP Inspector for local testing
- stdio transport for Claude Desktop / Cursor demos
Six-week journey
| Week | Focus |
|---|---|
| 1 | Set up & first MCP tool (greet ✅ in this repo) |
| 2 | Design your own tools → see docs/WEEK-2.md |
| 3 | Connect tools to real data |
| 4 | Make it safe & reliable |
| 5 | Test & write docs people can follow |
| 6 | Ship on GitHub & Demo Day |
Full program details: docs/PROGRAM.md
Starter project options (pick in Week 2)
- Notes & FAQ Search — fully offline (example stubs included)
- Personal Expense Tracker — summarize spending from a spreadsheet
- To-Do List — create / list / complete tasks
- Weather Briefing — free API (e.g. Open-Meteo), no paid keys
- Quote of the Day — simple offline or public API
Advanced ideas (repo health, course planner, job tracker) need mentor approval before you expand scope.
Repo layout after Week 2
MCPRepo/
├── docs/
│ ├── PROGRAM.md
│ ├── CURRICULUM.md
│ ├── WEEK-2.md
│ ├── project-choice.md
│ └── design.md
├── examples/
│ └── <tool_name>.json
├── src/
│ ├── index.ts
│ ├── schemas/
│ └── tools/
├── package.json
└── README.md
Rules that matter
- One job per tool; use
verb_nounnames (search_notes,add_expense) - Write descriptions for the model, not only for humans
- Every Zod field needs
.describe(...) - Prefer small focused tools over one mega-tool with an
actionenum - Avoid paid APIs / OAuth-heavy projects in Weeks 1–2
Links
- NextFlows Academy
- Program page (this repo)
- Apply for Cohort #1
- MCP docs
- MCP specification
- Build your first server (TypeScript SDK)
License
MIT — built for NextFlows Academy students.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.