product-search-mcp-server
An MCP server that exposes SF Supplies product search API to LLM tools. Supports free-text search, autocomplete suggestions, and retrieving filterable attributes.
README
product-search-mcp-server
An MCP server that exposes the SF Supplies product search (the Typesense-backed
website search API) to an LLM tool loop. It is a sibling of mssql-mcp-server
and uses the same transport/auth/deploy pattern, so the internal chatbot can
bridge both servers at once.
Tools
| Tool | What it does |
|---|---|
search_products |
Runs a search against POST /api/Products/Search/v2. Supports free-text query, attribute filters, category_slug, sorting, and pagination. Returns each product's name, page URL, stock status, and attributes (Brand, Color, Size, Material, …). |
suggest_products |
Fast typeahead/autocomplete (GET /api/Products/query?query=…). Returns matching product names + URLs only — lightweight. Use it to resolve a vague/partial term, then hand the chosen term to search_products. |
list_product_filters |
Returns the facets (filterable attributes and their values + counts) available for a query/category — so the model knows what it can pass to search_products' filters. |
The filters argument maps facet field names to arrays of values, matching the
API's attributeFilter, e.g.:
{"Brand": ["Avery"], "Color": ["Plum"], "Size": ["15\" x 10 Yards"]}
Run locally (stdio — e.g. Claude Desktop)
pipenv install
pipenv run python -m product_search_mcp_server
Run as a shared HTTP service (LAN)
cp .env.deploy.example .env.deploy # set MCP_AUTH_TOKEN (openssl rand -hex 32)
docker compose -f docker-compose.deploy.yml up -d --build
Consumers connect to http://<host>:8001/mcp with header
Authorization: Bearer <MCP_AUTH_TOKEN>. Defaults to port 8001 to avoid
clashing with the MSSQL MCP server on 8000.
examples/chatbot_bridge.py shows the client-side bridge the chatbot uses.
Configuration
| Env var | Default | Purpose |
|---|---|---|
MCP_TRANSPORT |
stdio |
stdio or http |
MCP_AUTH_TOKEN |
— | Required bearer token in http mode |
MCP_DISABLE_AUTH |
false |
Dev-only: serve HTTP with no auth |
MCP_HTTP_HOST / MCP_HTTP_PORT |
0.0.0.0 / 8000 |
HTTP bind (container) |
SEARCH_API_URL |
https://api.sfsupplies.com/api/Products/Search/v2 |
Results/filtering endpoint |
SUGGEST_API_URL |
https://api.sfsupplies.com/api/Products/query |
Typeahead/autocomplete endpoint |
PRODUCT_URL_TEMPLATE |
https://www.sfsupplies.com/product/{slug} |
Product page link template — verify the path |
SEARCH_API_TIMEOUT |
15 |
HTTP timeout (seconds) |
SEARCH_API_KEY / SEARCH_API_KEY_HEADER |
— | Optional, only if the API gets locked down |
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。