MCP App Store
by KeenEthics
PartFinder icon

PartFinder MCP Server for Claude and ChatGPT

By PartFinder

Overview

PartFinder gives Claude live supply chain data for electronic and industrial parts, so you can source and de-risk components in a conversation. Ask it to: - Research a part: who makes it, where it is made, stock and lead times across distributors, lifecycle, datasheet and a 0 to 100 supply risk score with the reasons behind it. - Find alternate parts, with their own stock and lead times. - Compare the specs of two to six parts side by side, from their datasheets. - Search for parts by part number or description. - Scan a bill of materials for risky lines. - Follow supply chain disruptions by country, region or category, and see where ports and shipping routes are running below their usual level. - Send a request for quote to a supplier and read the reply. You sign in with your PartFinder account and choose your organization. Reading data and making changes are separate permissions. Each tool call uses credits from your organization's PartFinder account, and admins can disconnect the app at any time in the PartFinder console.

How to connect

  1. Claude: Open the PartFinder listing in the Claude connectors directory and choose Connect. Claude listing PartFinder exposes 24 tools here.
  2. ChatGPT: Open the PartFinder listing in the ChatGPT app directory and choose Connect. ChatGPT listing PartFinder exposes 25 tools here.

Tools

analyzeBom

ChatGPT
Analyze a BOM. Submit a bill of materials for supply-chain risk analysis. Use it whenever your user asks to scan or analyze a BOM or a list of parts, even a list of one line. Every line is identified, its origin and current distributor availability resolved, and a risk score computed from manufacturing geography, live disruption exposure, lead-time volatility and sourcing gaps across the BOM. The work runs in the background: this call answers 202 at once with the BOM id, and getBom carries the report, partial while status is processing and full once it is complete. Time to complete varies with the parts in the BOM. One call carries up to 500 lines; a BOM is kept until you delete it. Then call getBom with the id at once: it returns when the BOM analysis finishes, or after 45 seconds with status processing; then call it again.

compareParts

ChatGPT
Compare parts. Put parts side by side on the specifications that matter for this kind of part, read from each part's datasheet. To compare parts you name with each other, send them all in parts, the first being the reference. Send mpn alone only to compare one part with its alternates. Every other part gets one verdict per specification against the reference (same, covers, narrower, higher, lower, different, not_stated) and a count of the specifications that are the same or covered, plus its stock and lead time at distributors. Answers 202 at once. A part that is not one specific part is refused with 422, a part no distributor lists with 404; the refusal names the part and nothing is started. Then call getComparison with the id at once: it returns when the comparison finishes, or after 45 seconds with status processing; then call it again.

deleteBom

ChatGPT
Delete a BOM. Removes the BOM and every line. Irreversible.

deleteComparison

ChatGPT
Delete a comparison. Removes the comparison and its results. Irreversible.

deleteRfq

ChatGPT
Delete a request for quote. Removes the request for quote and every message in its thread. Irreversible. A reply the supplier sends afterwards is not kept, and the supplier is not told.

deleteSearch

ChatGPT
Delete a search. Removes the search and its results. Irreversible.

getAlternates

ChatGPT
Find alternate parts. Parts that could replace one part, up to five, each with its current stock, distributor count, lead time and lifecycle. Never the part itself or the same part in another package. The list is kept server-side for 48 hours, so repeat calls return quickly; stock is read fresh on every call. Responses carry Cache-Control: no-store. Confirm fit and specifications against the datasheets before substituting.

getBom

ChatGPT
Get a BOM report. The report for one BOM: a risk average and distribution, the gaps across the BOM, where it is made, and every line with its own score, drivers and current supply. While bom.status is processing the lines already analyzed are scored and the rest are pending. Each poll counts toward your rate limit. Scores are computed at read time against current disruption events and distributor figures. While the BOM analysis runs, this tool waits: it returns when the BOM analysis finishes, or after 45 seconds with status processing; then call it again.

getComparison

ChatGPT
Get a comparison. A comparison and, once status is complete, its parts: the reference first, then each part compared with it. Each read counts as a call. failed means the comparison could not be completed: submit it again. While the comparison runs, this tool waits: it returns when the comparison finishes, or after 45 seconds with status processing; then call it again.

getEvent

ChatGPT
Get a disruption event. One event in full. The id is the one listEvents, researchPart and getBom return.

getRfq

ChatGPT
Get a request for quote. A request for quote and its email thread, oldest first: what was sent and every reply from the supplier. status is replied once the supplier has answered. Poll this to see new replies.

getSearch

ChatGPT
Get a search. A search and, once status is complete, the parts it found. Each read counts as a call. failed means the search could not be completed: submit it again. While the search runs, this tool waits: it returns when the search finishes, or after 45 seconds with status processing; then call it again.

listBomSnapshots

ChatGPT
List saved BOM reports. The reports saved for one BOM, newest day first: one for the day its analysis completed, then one a day, each scored as it was on that day. A BOM has one saved report per day. Each carries its day (taken_on), when it was saved, its risk average and band. Read one in full with getBom and as_of set to its taken_on. Saved reports are kept as long as the BOM. When next_cursor is not null, call again with cursor set to it for the next page.

listBoms

ChatGPT
List your BOMs. Every BOM your organization has submitted (with a key made with POST /v1/keys, only that customer's), newest first, with its status, one page at a time. When next_cursor is not null, call again with cursor set to it for the next page.

listComparisons

ChatGPT
List your comparisons. Every comparison your organization has run (with a key made with POST /v1/keys, only that customer's), newest first, one page at a time. When next_cursor is not null, call again with cursor set to it for the next page. Read one with getComparison for its parts.

listEventCountries

ChatGPT
Disruptions by country. The countries on the risk map on partfinder.dev right now: recent events still unfolding, how many, and the worst severity among them. It is not every country with an active event: a country absent here can still have active events, listed by listEvents with country set. Open the most severe of a country's listed events with getEvent and top_event_id.

listEvents

ChatGPT
List disruption events. Active supply-chain disruption events, newest first, one page at a time: what happened, where, how serious, which component categories it affects, what it means for a buyer, and the news reports of it. Filter by country, region, category or how recently it was first recorded. When next_cursor is not null, call again with cursor set to it for the next page.

listRfqs

ChatGPT
List your requests for quote. Every request for quote your organization has sent (with a key made with POST /v1/keys, only that customer's), newest first, with its status, one page at a time. When next_cursor is not null, call again with cursor set to it for the next page. Read one with getRfq for its messages.

listSearches

ChatGPT
List your searches. Every search your organization has run (with a key made with POST /v1/keys, only that customer's), newest first, one page at a time. When next_cursor is not null, call again with cursor set to it for the next page. Read one with getSearch for its rows.

listShippingConditions

ChatGPT
Port and sea-route activity. Port calls per country and ship transits per sea route over the latest week, as a change in percent against their usual level for this time of year. A drop at a port or a strait delays shipments through it.

ping

ChatGPT
Verify an API key. Returns the organization the key belongs to. Use it to confirm an integration is wired correctly before calling an endpoint that does real work.

researchPart

ChatGPT
Research a part. Supply-chain intelligence for one industrial part: a composite risk score with its four drivers, current distributor availability, a datasheet link, the live disruption events that touch it, alternate parts, and recommended actions. A cold part takes tens of seconds because it gathers live data. The result is kept server-side for 48 hours, so repeat calls return quickly; responses themselves are not HTTP-cacheable and carry Cache-Control: no-store.

searchParts

ChatGPT
Search for parts. Find parts for a part number or a description: the part itself and comparable parts from the same or other makers, each with its product page, datasheet, the specifications that matter for this kind of part, and its stock and lead time at distributors. Answers 202 at once. A query that is not about a part is refused with 422. Then call getSearch with the id at once: it returns when the search finishes, or after 45 seconds with status processing; then call it again.

sendRfq

ChatGPT
Send a request for quote. Email a request for quote to a supplier you name, under the requester's name. This sends a real email to supplier.email: every address and name in the request must come from your user, never be guessed or suggested; ask your user for any that are missing. Only these are required: the supplier's email, the requester's name, and the requester's company when it is neither your organization nor, with a key made with POST /v1/keys, that customer. Everything else is optional: leave out what they did not give. Send message only with text your user wrote; omit it and an email is written for you. The supplier replies to the email as usual, and their replies appear on getRfq. A supplier who asked not to receive requests for quote is refused with 409. This tool sends an email to the supplier named in the request.

sendRfqMessage

ChatGPT
Send a message in a request for quote. Answer the supplier, or send a reminder, in the same email thread as the request for quote. The text is sent as written, under the requester's name. This tool sends an email to the supplier named in the request.

analyzeBom

Claude

compareParts

Claude

deleteBom

Claude

deleteComparison

Claude

deleteRfq

Claude

deleteSearch

Claude

getAlternates

Claude

getBom

Claude

getComparison

Claude

getEvent

Claude

getRfq

Claude

getSearch

Claude

listBoms

Claude

listComparisons

Claude

listEventCountries

Claude

listEvents

Claude

listRfqs

Claude

listSearches

Claude

listShippingConditions

Claude

ping

Claude

researchPart

Claude

searchParts

Claude

sendRfq

Claude

sendRfqMessage

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/partfinder"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/partfinder)