About porefind
porefind is a local bridge to Filtrous’ hosted catalogue of laboratory filtration products, for researchers and lab purchasers. It supports specification-based product discovery and detail lookup, with current price and lead-time queries available through an API-keyed tool.
Key features
- 01Search products by text and specifications
- 02View product variants and specifications
- 03Check SKU price and availability
- 04Retrieve lead time for a SKU
Use cases
- 01Find filters by pore size or material
- 02Compare product options and specifications
- 03Check current pricing and lead time for a SKU
MCP support
It exposes three read-only Filtrous catalogue tools through a local MCP server, forwarding calls to the hosted endpoint. The quote-request tool is omitted.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
3Saved tool list · updated 1w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| limit | integer | No | Max results (default 10). |
| query | string | Yes | Free-text and/or spec search. Field filters like 'product_type:...' or 'tag:...' are also honored. |
Details
From the project READMEporefind
porefind is a thin local bridge to Filtrous' hosted catalog of laboratory filtration products: syringe filters, membranes, filter paper and the like, each with its pore size, material, SKU, price and stock position. The bridge stores nothing. It presents the upstream server's tool list locally, exactly as captured, and passes each tool call through to the remote endpoint, handing back whatever comes back untouched.
Catalog work here is read-only. You can search the catalog by free text or by specification terms, open one product for its full detail — variants, options, images and spec metafields — and ask for the current price and lead time on a single SKU. Pricing for the last of these sits behind the upstream's API key, as its own description says.
Labelled subset: 3 of 4 tools
The live capture of the upstream server listed four
tools. This build serves three. The fourth,
request_quote, took a list of items plus a
contact object and submitted them to the vendor as
a quote request — personal contact details routed
into a sales pipeline. That is out of scope for this
program under its safety rule, so the tool is
omitted entirely: it is not registered, it does not
appear in tools/list, calls to it are rejected like
any unknown name, and the test suite asserts its
absence at both the registry and the served MCP
layer.
Tools
search_products— search the catalog by free text and/or spec terms (for example0.22 micron PES syringe filter); field filters such asproduct_type:...ortag:...are also honoured. Returns matching products with title, handle, SKU, product type, tags, price and availability. Required:query. Optional:limit(1-50, default 10).get_product_details— full detail for one product by its handle (the slug a search returns): variants, SKUs, options, prices, images and spec metafields. Required:handle.get_price_and_lead_time— current price, availability and, when permitted, stock for one exact SKU. Required:sku. This is the keyed tool: it needs a valid API key upstream.
Configuration
FILTROUS_MCP_URL(optional) — override the upstream endpoint. Defaults tohttps://filtrous-mcp-catalog.filtrous.workers.dev/mcp. (The override hook is this rebuild's own labelled reconstruction.)FILTROUS_API_KEY(optional) — when set, every upstream request carries anAuthorization: Bearer <key>header; when it is not set, no auth header is sent. (This forwarding is also a labelled reconstruction; the original surface was captured keyless.)
Forwarding mechanics
The first tool call opens a streamable HTTP session
with the upstream: a JSON-RPC initialize request
(protocolVersion 2025-03-26, clientInfo name
porefind), then a notifications/initialized
notification, then the tools/call request carrying
the caller's arguments unchanged. If the upstream's
initialize reply carries an Mcp-Session-Id
header, that value rides on every later request of
the session, which is reused for the life of the
process. Replies may arrive as plain JSON or as
server-sent events; both are read.
Successful upstream results pass through verbatim,
including any isError flag the upstream sets.
Failures — a JSON-RPC error, a non-2xx status, an
unreadable response, or a request that cannot be
completed — come back as an isError result whose
text starts with porefind error: (this rendering
is this rebuild's own labelled reconstruction).
CLI
Each kept tool is also a subcommand taking one JSON argument string:
npx @reneturcios/porefind search_products '{"query":"0.22 micron PES syringe filter","limit":5}'
npx @reneturcios/porefind get_product_details '{"handle":"pes-syringe-filter-022"}'
npx @reneturcios/porefind get_price_and_lead_time '{"sku":"SF-PES-022-25"}'
npx @reneturcios/porefind --version
Exit codes: 0 on success, 1 on an isError
result, 2 on a usage error.
MCP server
npx -y @reneturcios/porefind mcp
MCP client config:
{
"mcpServers": {
"porefind": {
"command": "npx",
"args": ["-y", "@reneturcios/porefind", "mcp"]
}
}
}
Docker
docker run -i ghcr.io/cloudcorprecords/porefind
docker run --rm ghcr.io/cloudcorprecords/porefind --version
The container's default command is mcp, so the
first form is a stdio MCP server; any other
arguments reach the CLI normally.
Development
npm install
npm run build
npm test
npm run dev
The test suite runs entirely against a stubbed JSON-RPC upstream — no live network calls.
License
MIT — see LICENSE.
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.