Connect Your Data Starter
Let AI assistants like Claude and Cursor safely use your own data or tools, starting from a working, tested template.
Your first result: your AI assistant can search a sample notes store through your own server.
$39
One-time ยท instant download
Coming soon. Leave your email and we'll tell you the day it opens.

What you get
What's inside the kit
A ready-to-run server in TypeScript
A working starting point you can rename and point at your own API.
The same server in Python
Pick the language your team already uses.
Safe defaults
Read-only by default; the one write tool is off unless you switch it on, and asks for confirmation.
Copy-and-paste setups for Claude Desktop, Claude Code and Cursor
Connect your server to the assistant you already use.
Tests you can run
Check everything works before you change it, and keep it working as you go.
Security and packaging guides
How to add a token for local HTTP access and how to package your server.
How to put it to work
How to get started
Follow these steps in order. After you buy, the same steps appear as a checklist in your kit viewer, with links to the exact guide for each step.
- 1
Pick your language and build
10 minIn the typescript (or python) folder, install the exact versions and run the tests.
- 2
Try it with the free testing tool (MCP Inspector)
5 minList the tools and call search_notes from the command line to see a response.
- 3
Connect your AI assistant
10 minUse the matching client config for Claude Desktop, Claude Code or Cursor and replace the paths with your own. These configs haven't been tested inside the real apps yet, so expect to adjust them.
- 4
Ask it something
1 minAsk "search my notes for welcome". Your assistant should show a search_notes tool call.
- 5
Make it yours
variesReplace the sample notes store with calls to your own API, keep inputs strict and add tests.
Is it right for me?
Who it's for
A great fit ifโฆ
- Developers and agencies connecting internal tools, APIs or data to AI assistants
- Comfortable with TypeScript or Python
Probably not for you ifโฆ
- You need sign-in (OAuth) built for you: guides only, no OAuth code
Questions
Frequently asked questions
Is it secure out of the box?
It starts with safer defaults (read-only, localhost-only, token for HTTP), but it isn't a security audit. You're responsible for securing your own deployment.
Does it include OAuth sign-in?
No. There's an auth guide but no OAuth code.
Has it been tested inside Claude and Cursor?
It was tested locally with the SDK client and the MCP Inspector. It hasn't been tested inside the Claude or Cursor apps yet.
Good to know before you buy
- Provided as is, without warranty; liability limited to the price you paid.
- No promise of any particular outcome, time savings or income.
- Not a security audit. You are responsible for securing your own deployment (auth, TLS, network exposure, secrets, access control).
- Exposing a server beyond localhost or enabling write tools is at your own risk.
- Log redaction is pattern-based and can miss secrets in unusual formats.
Technical details (for the technically curious)
Technical words on this page, explained
- MCP:
- Model Context Protocol: a common plug that lets AI apps like Claude or Cursor connect to your data and tools.
- OAuth / sign-in client:
- The safe "Sign in with Google" style of connecting an app, where you choose exactly what it may access.
- CLI / terminal:
- Typing short commands into a text window on your computer instead of clicking buttons.
- CI / GitHub workflow:
- Automatic checks that run every time code changes, so problems are caught early.
- Env file / secrets:
- Private settings like passwords and keys. They must never be shared or saved in public code.
Formerly / technical name: MCP Server Starter Kit (TypeScript + Python)
Full description
MCP server starter templates in TypeScript and Python with safer defaults (read-only tools, localhost-only token-auth HTTP), stdio and Streamable HTTP transports, tests and packaging guides. OAuth is not included. Not a security audit; you secure your deployment.
Full parts list
- TypeScript server (@modelcontextprotocol/sdk 1.32.2)
- Python server (mcp 2.3.0)
- Read tools + confirm-gated write tool (off by default)
- stdio and Streamable HTTP (bearer token, 127.0.0.1)
- Unit + end-to-end tests incl. MCP Inspector CLI
- Client configs for Claude Desktop, Claude Code, Cursor
- Auth and packaging guides (no OAuth code)
- SHA-pinned CI workflow
Requirements
- Node 20+ and/or Python 3.11+
- An MCP client
Tested on
- Node 20.19.2 / npm 11.21.0, TypeScript 5.9.3
- @modelcontextprotocol/sdk 1.32.2
- Python 3.13.5 and 3.11.17, mcp 2.3.0
- uv 0.13.1, Linux 6.12.94+
Tested locally (SDK client + MCP Inspector CLI). Not yet tested inside Claude or Cursor apps; CI not run on GitHub. (2026-10-28)
What's in the download (44 files)
- ๐ .github/ (1 file)
- ๐ client-configs/ (3 files)
- ๐ docs/ (2 files)
- ๐ python/ (14 files)
- ๐ scripts/ (1 file)
- ๐ tests/ (1 file)
- ๐ typescript/ (14 files)
- ๐ .gitignore
- ๐ CHANGELOG.md
- ๐ LICENSE
- ๐ package.json
- ๐ README.md
- ๐ REVIEW_STATUS.md
- ๐ SETUP.md
- ๐ THIRD_PARTY_NOTICES.md
Version and changes
v0.1.0 ยท tested 2026-10-28
- 0.1.0 ยท 2026-10-28: Initial release: TS and Python servers, transports, auth, tests, client configs, pinned CI. Compliance review: license, notices, disclaimers.