About Cellarion
Cellarion is an MCP bridge to the Cellarion wine cellar management service, for people using stdio-based MCP clients to access their personal cellar data. It runs locally, connects with a personal token, and gets its tools and resources from the server.
Key features
- 01Check bottles and storage locations
- 02View drinking windows
- 03Review drinking history and ratings
- 04Fetch tools and resources from the server
- 05Use revocable read-only tokens
Use cases
- 01Decide which bottles to open soon
- 02Find a wine and its storage location
- 03Review last month's wines and ratings
MCP support
The bridge connects stdio-based MCP clients to Cellarion's /api/mcp endpoint and proxies the tools and resources defined by the server.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://cellarion.app/api/mcpclaude mcp add --transport http app-cellarion-cellarion https://cellarion.app/api/mcp{"mcpServers":{"app-cellarion-cellarion":{"type":"http","url":"https://cellarion.app/api/mcp"}}}Open in Cursor{"servers":{"app-cellarion-cellarion":{"type":"http","url":"https://cellarion.app/api/mcp"}}}Open in VS Code{"mcpServers":{"app-cellarion-cellarion":{"type":"http","url":"https://cellarion.app/api/mcp"}}}{"mcpServers":{"app-cellarion-cellarion":{"serverUrl":"https://cellarion.app/api/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEcellarion-mcp
Connect your AI to your wine cellar. This is the MCP bridge for Cellarion — it lets stdio-based MCP clients (Claude Desktop, Cursor, …) talk to your Cellarion account, so you can ask your assistant things like:
- "What should I open tonight before it goes past its peak?"
- "Do I still have any Barolo, and where is it stored?"
- "What did I drink last month, and what did I rate it?"
It runs entirely on your machine and proxies to your Cellarion server's /api/mcp endpoint using a personal, revocable API token. Tools and resources are defined by the server — updating Cellarion automatically updates what your AI can do; this bridge never needs to know.
Setup
- Mint a token in Cellarion: Settings → API tokens → create a token with the
readscope. Copy thecel_…value (shown once). - Add the server to your client. Claude Desktop (
claude_desktop_config.json):
{
"mcpServers": {
"cellarion": {
"command": "npx",
"args": ["-y", "cellarion-mcp"],
"env": {
"CELLARION_TOKEN": "cel_your_token_here"
}
}
}
}
- Restart the client. Ask it about your cellar.
Self-hosting Cellarion? Point the bridge at your instance:
"env": {
"CELLARION_TOKEN": "cel_your_token_here",
"CELLARION_URL": "https://cellarion.your-domain.example"
}
(--token / --url command-line arguments work too.)
Security notes
- The token is scoped and revocable instantly from Cellarion — revoking it cuts this bridge off mid-session.
- A
read-scope token can only read your own cellar data. It cannot modify anything, cannot read other accounts, and cannot manage tokens. - The token goes only to the host in
CELLARION_URL— nowhere else. Nothing is logged. - If the connection drops mid-request, the bridge reconnects. Tool and resource lists are fetched again, and so is a call to a tool the server lists as read-only. A call to any other tool (which may change your cellar) is sent again only when it carries an
idempotency_key, because the server then returns the first result instead of acting twice. Otherwise your AI gets the error and can check what happened before trying again. A request the server turned away because the bridge's session had expired never ran, so the bridge opens a new session and sends it again.
License
AGPL-3.0, like Cellarion itself. Source: https://github.com/jagduvi1/Cellarion (cellarion-mcp/).
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.