Pantrist

Pantrist

Manage your Pantrist pantry, shopping list, recipes, and weekly meal plan from Claude. The server wraps the same public REST API that powers the iOS, Android, and web apps, so your changes show up in real time on every device. * "What's on my shopping list?" * "Add milk, eggs, and that brand of olive oil I had last week" * "What's expiring in the next three days?"

Category
Visit Server

README

Pantrist MCP Server

A Model Context Protocol server that wraps the Pantrist REST API, so an LLM client (Claude Desktop, the Claude web/mobile connector, Cursor, …) can manage shopping lists, the pantry, recipes and the week plan in natural language.

It's a thin wrapper — no business logic. Every tool maps to an existing REST endpoint and forwards the caller's Bearer token. The HTTP client is generated from the public OpenAPI spec (src/generated/pantrist-api.ts), so request/response types track the API automatically; only the curated tool layer (src/tools.ts) is hand-written.

Documentation

  • docs/ARCHITECTURE.md — components, the token context, request flow, transports, regeneration.
  • docs/AUTHENTICATION.md — OAuth flow, token-type ↔ tool compatibility, multi-user isolation, the consent-page dependency.
  • docs/DEPLOYMENT.md — full env var reference, remote/ingress setup, scaling, security checklist.
  • docs/TOOLS.md — every tool's args, REST mapping, and the item shape.
  • docs/LIMITATIONS.md — known rough edges (read before relying on it in production).

Two transports

Transport When Auth
stdio (src/stdio.ts) Local PoC, single user, Claude Desktop Bearer from PANTRIST_TOKEN env
Streamable HTTP (src/http.ts) Remote, multi-user, the Claude connector Per-request Bearer, obtained by the client via OAuth

Connecting Claude to the hosted server

If you just want to use Pantrist with Claude (not self-host), the public endpoint is https://mcp.pantrist.app/mcp. Pick whichever Claude surface you're on; in every case the first tool call walks you through an OAuth login to your Pantrist account, no token to copy by hand.

Claude.ai (web or desktop app)

  1. claude.ai → profile menu → Settings → Connectors
  2. Click Add custom connector
  3. Fill in:
    • Name: Pantrist
    • URL: https://mcp.pantrist.app/mcp
  4. Save → click Connect. A popup opens the Pantrist consent page.
  5. Sign in → authorize → the popup closes.
  6. Start a new chat — the Pantrist tools (shopping list, pantry, week plan, recipes) show up in the tool selector. Try "What's on my shopping list?"

Claude Code (CLI)

claude mcp add pantrist --transport http https://mcp.pantrist.app/mcp

In the session, run /mcp to confirm it's listed. The first tool call triggers the OAuth flow in your browser.

Claude Desktop (manual config)

For Claude Desktop versions that support remote MCP, in ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on Windows/Linux:

{
  "mcpServers": {
    "pantrist": {
      "type": "url",
      "url": "https://mcp.pantrist.app/mcp"
    }
  }
}

Restart Claude Desktop. First tool use kicks off OAuth.

Sanity-checks if it doesn't work

# 200 OK + {"status":"ok"}
curl -fsS https://mcp.pantrist.app/healthz

# 401 + a WWW-Authenticate header pointing at /.well-known/oauth-protected-resource
curl -i -X POST https://mcp.pantrist.app/mcp -H 'Content-Type: application/json' -d '{}'

# JSON listing api.pantrist.app as the authorization server
curl -fsS https://mcp.pantrist.app/.well-known/oauth-protected-resource

If all three succeed but the connector flow still fails, the API's OAUTH_AUTHORIZE_URL probably isn't set to a browser-facing consent page — see the warning in the OAuth dependency note.

Quick start (stdio — fastest path)

You can validate the whole tool set in a couple of minutes without touching OAuth, by using a token you already have.

git clone https://github.com/NLueg/pantrist-mcp.git
cd pantrist-mcp
npm install
npm run build

# Get a token: log into the Pantrist app, open DevTools → Network, copy the
# `Authorization: Bearer <…>` value from any API request (a Firebase ID token —
# the API accepts it directly). Or use an OAuth access_token.
export PANTRIST_BASE_URL=https://api.pantrist.app
export PANTRIST_TOKEN=<paste-token>
export PANTRIST_LIST_ID=<a-list-uuid>   # optional; or call list_lists

npm run dev:stdio   # or: node dist/stdio.js

Claude Desktop config

claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "pantrist": {
      "command": "node",
      "args": ["/ABSOLUTE/PATH/pantrist-mcp/dist/stdio.js"],
      "env": {
        "PANTRIST_BASE_URL": "https://api.pantrist.app",
        "PANTRIST_TOKEN": "<your-token>",
        "PANTRIST_LIST_ID": "<list-uuid>"
      }
    }
  }
}

Restart Claude Desktop, then try: "What's on my shopping list?" or "Add milk and eggs."

Remote (Streamable HTTP + OAuth)

export PANTRIST_BASE_URL=https://api.pantrist.app
export MCP_PUBLIC_URL=https://mcp.pantrist.app   # public URL of THIS server
export MCP_ALLOWED_HOSTS=mcp.pantrist.app        # optional DNS-rebinding guard
export PORT=8787
npm run dev:http   # or: node dist/http.js

Then add it in Claude as a Custom Connector with URL https://mcp.pantrist.app/mcp. The server also exposes GET /healthz for probes. Full env reference, ingress, and scaling notes are in docs/DEPLOYMENT.md.

How the OAuth handshake flows

Claude ──POST /mcp (no token)──▶ MCP server
        ◀── 401 + WWW-Authenticate: resource_metadata=".../oauth-protected-resource"
Claude ──GET  /.well-known/oauth-protected-resource ──▶ MCP server
        ◀── { authorization_servers: ["https://api.pantrist.app"] }
Claude ──GET  /.well-known/oauth-authorization-server ─▶ pantrist-api   (RFC 8414)
Claude ──POST /access-token/register ─────────────────▶ pantrist-api   (RFC 7591 DCR)
Claude ──(browser) authorization_endpoint ────────────▶ consent page   (see below)
Claude ──POST /access-token/token (code + PKCE) ──────▶ pantrist-api   → access_token
Claude ──POST /mcp (Bearer access_token) ─────────────▶ MCP server ──▶ REST API

The MCP server is the Resource Server; the Authorization Server is the Pantrist API. The token Claude receives is the API Bearer, so this server just forwards it.

⚠️ Dependency — the consent page. The API's authorization_endpoint must be a browser-navigable login/consent page (the API's /access-token/authorize is a guarded JSON endpoint and can't be navigated to directly). Host one on the app (e.g. https://pantrist.app/oauth/authorize) and set the API's OAUTH_AUTHORIZE_URL env to point at it. Until that page exists, use the stdio path above with a manually-supplied token.

Tools

Tool REST route
list_lists GET /list
list_shopping_items GET /list/{listId}/shoppingList
add_shopping_item POST /list/{listId}/shoppingList/add-by-name
check_shopping_item POST /list/{listId}/shoppingList/{itemId}/check
delete_shopping_item DELETE /list/{listId}/shoppingList/{itemId}
list_pantry_items GET /list/{listId}/pantryList
add_pantry_item POST /list/{listId}/pantryList/add-by-name
reduce_pantry_amount PUT /list/{listId}/pantryList/{itemId}/change-amount
update_pantry_item GET + PUT /list/{listId}/pantryList/{itemId} (metadata-only; stock changes go through reduce_pantry_amount)
search_recipes POST /recipe/filter
get_recipe GET /recipe/{recipeId}
delete_recipe DELETE /recipe/{recipeId}
get_week_plan GET /list/{listId}/weekPlan?from=&to=
update_week_plan_day PUT /list/{listId}/weekPlan/{date}

Most tools accept an optional listId; if omitted they use PANTRIST_LIST_ID in stdio mode only (HTTP mode requires it explicitly — see multi-user isolation). Full argument and item-shape details are in docs/TOOLS.md.

All of these are public API endpoints (present in /swagger-ui-json), so this wrapper needs only the published spec — never the private API source. That keeps the door open to open-sourcing this directory as its own repo.

Tests

npm test     # Node's built-in test runner (via tsx) — wiring + multi-user gating

Regenerating the API client

Two steps, run when the API contract changes:

# 1. In the pantrist-api repo: emit the public OpenAPI spec
#    (Nest preview mode — no DB). Writes the snapshot directly into
#    ../pantrist-mcp/openapi/pantrist-openapi.json.
cd ../pantrist-api && pnpm generate:openapi

# 2. Back here: regenerate the typed client from that spec.
cd ../pantrist-mcp && npm run generate:client

Both the spec snapshot (openapi/pantrist-openapi.json) and the generated client (src/generated/pantrist-api.ts) are committed so the project builds without network access. The tool layer in src/tools.ts is hand-authored and not regenerated.

Environment

See .env.example.

Recommended Servers

playwright-mcp

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.

Official
Featured
TypeScript
Magic Component Platform (MCP)

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.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

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.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

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.

Official
Featured
TypeScript
Kagi MCP Server

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.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

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.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured