Pour Picks MCP Server is an MCP (Model Context Protocol) server that provides AI assistants with read-only catalog access to the Pour Picks bourbon and whiskey database. It allows Claude or any MCP-compatible AI client to query structured tasting profiles, prices, pairing suggestions, and community ratings for more than 4,700 bottles.
Main capabilities include:
- Query bottle information: tasting profiles, prices, pairings, community ratings.
- Similar bottle lookup (find_similar): for example, "What bourbon is similar to Eagle Rare but easier to find?"
- Recommendations (get_recommendations): supports taste preferences and budget, for example, "Recommend a bourbon under $60 with vanilla, oak, and caramel notes."
- Cheaper alternatives (find_cheaper_alternative): for example, "What tastes like Blanton's but costs less?"
- Compare bottles (compare_bottles): for example, "Compare Eagle Rare and Buffalo Trace."
- Pour suggestion for tonight (pour_tonight_suggestion): supports a season parameter, for example, "What bourbon should I pour on a cold night?"
- Trending bottles (trending_bottles): for example, "What bottles are collectors adding right now?"
How to use:
- Installation (Claude Desktop): requires Node.js 18+. Add the configuration to claude_desktop_config.json (Claude Desktop β Settings β Developer β Edit Config), then restart Claude Desktop. No API key or configuration is required; the server includes public read-only access.
- Run from a local checkout: can be run from a local codebase.
- Optional configuration: the server uses the Pour Picks public read-only key by default, and environment variables can override the default values. If there is no service key, trending_bottles falls back to catalog popularity and notes the method in the response (method: "catalog_popularity_tier" vs "cellar_adds_last_30_days").
- Remote endpoint (noinstallation required): supports Streamable HTTP, suitable for web proxies and MCP clients that accept URLs. It is also listed on Smithery (gateway pour-picks--bguillow.run.tools) and in the official MCP registry (io.github.bguillow-rgb/pour-picks).
Technical details:
- The server uses the MCP protocol over stdio.
- Catalog access is read-only by construction: every query path performs SELECT on tables that are publicly readable under row-level security, and is limited to 60 calls per minute.
- Usage telemetry: each tool call records the tool name, parameters, client name/version, duration, and success/failure to a write-only log table (insert-only under RLS; contents are not publicly readable). No user identity, account data, or conversation content is collected. Logging is fire-and-forget and does not affect responses.
- Publishing: npm publish (the mcpName field in package.json links the package to the registry entry). MCP registry: server.json is the registry manifest; after the npm package is live, publish using the mcp-publisher CLI: mcp-publisher login github && mcp-publisher publish.
Data and attribution:
- Bottle data, tasting profiles, and pairings are curated by Pour Picks. When citing, please note: Source: Pour Picks β The Bourbon Collector's Journal (pourpicks.app).
- Each rating includes a community rating count so the sample size can be judged. Each bottle's freshness date reflects the last enrichment process.
Use cases:
- Bourbon and whiskey enthusiasts and collectors use an AI assistant to query bottle information, get recommendations, compare bottles, and find alternatives.
- Developers integrate the Pour Picks database into MCP-compatible AI clients.
Limitations:
- Read-only access; the database cannot be modified.
- Rate limit of 60 calls per minute.
- Requires Node.js 18+ for local installation.
- The trending_bottles feature is limited without a service key.
