The reason there are separate CPU, GPU, RAM, SSD and motherboard tools is that the useful filters differ per part — socket and cores for a processor, chipset and memory for a graphics card, interface and capacity for a drive. That is what makes "an AM5 board with a B650 chipset under this budget" answerable. The data is a local snapshot, so its freshness is entirely down to when you last ran the parser.
A server over a locally parsed snapshot of the CoolPC price list. A Python parser downloads and converts the site's listing into product.json, and the MCP server queries that file, so searches are fast and offline.
- search_products filters by keyword, category and a minimum and maximum price
- get_product_by_model returns one product by model number, and get_category_products returns everything in a category
- list_categories enumerates the categories with counts
- search_cpu filters by socket and core count with price sorting; search_gpu filters by chipset and memory size
- search_ram filters by type, total capacity and frequency; search_ssd filters by interface and capacity
- search_motherboard filters by socket, chipset and form factor
- The component searches sort by price ascending or descending and return 10 results by default
- Listings carry the site's markers for best sellers, price drops and limited-time offers
Python 3.x for the parser and Node.js 18 or higher for the server. Generate the data first — `python3 coolpc_parser.py --download --json product.json` — then build with `npm run build` and point the client at build/index.js. Prices are only as current as your last parser run, so re-run it before quoting anything.