The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
medical-terminologies-mcp is a Model Context Protocol server that lets an MCP-compatible AI assistant run structured lookups against medical code systems. The walkthrough that introduced it, published by Sidney Bissoli on DEV Community on May 12, 2026, frames it as a seven-system tool: ICD-11, LOINC, RxNorm, MeSH, ATC, Brazilian Portuguese CID-10, and optional SNOMED CT. Instead of asking a model to recall a code from memory, the assistant calls the server, which searches the named source and returns candidate codes with their labels and attributes.
The distinction that matters is between looking up a terminology entry and making a clinical decision. The server answers “what is this code, and what does its record say?” It does not answer “what should this patient receive?” The rest of this article works through the setup, the three demonstrations in the walkthrough, and the checks needed before the tool goes near a clinical or research workflow.
What the server does and does not do
- Searches terminology sources. A client sends a term, such as a lab test name or drug name, and the server returns matching entries from the named systems.
- Shows detail records. A second lookup returns the attributes of a specific entry, including whether it is active. The walkthrough treats this inspection step as essential before a code is used.
- Maps between systems, with different levels of support. Some crosswalks are backed by bundled WHO transition tables; others are guidance only. The mapping section below explains the difference.
- Offers additional features. The walkthrough also describes batch code validation, terminology version and difference information, MCP Prompts and Resources, structured output, API rate limits, retries, and two transports. These are project-author descriptions; they have not been independently tested.
The server does not diagnose, does not order or prescribe, and does not confirm that a literature search is complete.
The seven terminology sources and how each is reached
The table below combines what the walkthrough and the project repository say about each system. Where either source is silent on how a system is accessed, the cell says so rather than guessing.
#1 Best Overall
| System | Source named | Access in the described setup |
|---|---|---|
| ICD-11 | WHO | Live API; WHO credentials are required for live lookup |
| ICD-10 to ICD-11 mapping | WHO transition tables (2025-01) | Bundled with the server |
| LOINC | Not stated in the walkthrough | Not stated which functions use a live API |
| RxNorm | Not stated in the walkthrough | Not stated which functions use a live API |
| MeSH | NLM | Not stated which functions use a live API |
| ATC | NLM RxClass | Not stated which functions use a live API |
| CID-10 (Brazilian Portuguese) | DataSUS, version V2008 | Bundled dataset, no live call needed |
| SNOMED CT (optional) | SNOMED International | Requires a SNOMED International license and a self-hosted Snowstorm instance |
CID-10 is the Portuguese-language name for ICD-10, which is why the server lists it as a separate system. The walkthrough states that WHO credentials are needed only for live ICD-11 lookup; in its described configuration, the other 26 functions do not require credentials.
Setting up the server
The walkthrough runs the server through npx, which requires Node.js to be installed on the machine. The following is the standard MCP client entry for a local server. The exact file location and reload step depend on your client, so follow its own documentation.
- Confirm that Node.js and npx are available from your terminal.
- Open your MCP client’s server configuration file and add an entry named
medical-terminologies. - Set the entry to run the package with npx:
{
"mcpServers": {
"medical-terminologies": {
"command": "npx",
"args": ["-y", "medical-terminologies-mcp"]
}
}
}- Save the file and restart or reload the client as its documentation describes.
- If you need live ICD-11 lookup, supply your WHO credentials through the environment settings described in the package documentation. Other functions do not need them.
The walkthrough also identifies a hosted endpoint that uses the Streamable HTTP transport, in addition to the local stdio setup shown above. Hosted availability and its terms can change, so confirm them with the project before relying on it.
Walkthrough 1: Confirming a LOINC code for a lab order
The first example asks: “What’s the LOINC code for procalcitonin in serum?” It shows a two-step pattern that is useful for any terminology lookup.
- Ask the assistant the question in plain language. The server runs a search and returns candidate entries.
- In the walkthrough, the search returns
33959-8, labeled “Procalcitonin [Mass/volume] in Serum or Plasma.” - Request the detail record for that code. The walkthrough uses this to display the entry’s attributes and whether it is active.
- Before the code is entered into an order set, a result system, or a data extract, confirm it against your laboratory’s own catalogue. An inactive or mismatched entry should not be used.
The walkthrough also gives lactate (2524-7), high-sensitivity troponin I (67151-1), and SpO₂ (59408-5) as further examples. Treat these as the walkthrough’s examples and confirm each one in the LOINC database before relying on it.
Walkthrough 2: Breaking down a combination drug
The second example comes from a medication reconciliation question: “I need to break down Janumet for the admission med rec. What are the active ingredients and their drug classes?” The assistant searches for Janumet, retrieves its ingredients, and then looks up the ATC classification for each one.
| Ingredient | ATC code shown in the walkthrough | ATC group name (WHO ATC classification) |
|---|---|---|
| Sitagliptin | A10BH | Dipeptidyl peptidase 4 (DPP-4) inhibitors |
| Metformin | A10BA and A10BD | Biguanides (A10BA); combinations of oral blood glucose lowering drugs (A10BD) |
This is an information-retrieval workflow. The output shows which ingredients and classes a combination product is recorded under. It does not, on its own, establish what a reconciliation requires, whether an interaction matters for a given patient, or what should be prescribed. Those decisions belong to the clinician and the pharmacy team, working from the patient’s record and the product’s full labeling.
Walkthrough 3: Building a MeSH-aware search strategy
The third example supports a researcher drafting a systematic review on primary health care interventions and avoidable hospitalizations. The walkthrough uses the server to make controlled vocabulary visible while the search is being built.
- Search the MeSH descriptor for Primary Health Care.
- Read the descriptor’s tree positions, which show where the term sits in the MeSH hierarchy.
- Review the allowed qualifiers. The walkthrough selects
/utilizationand/statistics & numerical datawhen drafting the PubMed query. - Check each added qualifier against the question being asked before running the search.
The benefit is transparency: the researcher can see which vocabulary terms and subheadings are being used and why. The server does not replace expertise in review methods, does not decide which subject headings best capture the question, and does not confirm that the resulting search is complete. A librarian or information specialist should still test the strategy against known relevant articles.
Mapping: what is table-backed and what is guidance only
Mapping support is uneven, and the walkthrough is explicit about the difference.
| Mapping | Basis described in the walkthrough | Status |
|---|---|---|
| ICD-10 to ICD-11 | Bundled WHO transition tables, 2025-01 | Table-backed. The walkthrough reports 11,243 ICD-10 categories (WHO 2025-01 transition tables, as reported by Sidney Bissoli), of which 1,461 have multiple WHO-documented ICD-11 candidates (WHO transition-table data, as reported by Bissoli). |
| LOINC to SNOMED CT | Guidance only | Not an authoritative curated mapping in the described version |
| SNOMED CT to ICD-10 | Guidance only | Not an authoritative curated mapping in the described version |
The two counts are the walkthrough’s figures, not values checked here against WHO’s published tables. Where a category has several candidates, the choice among them is a judgment that the server does not make for you.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteVersions and counts: what changed after May 2026
The walkthrough describes a seven-system tool with 31 tools. Later project materials differ, so the figures below should be read by source and date.
Best Value
- Essential guide to the language of medicine
- Includes 1 000 new words and senses
- Covers the latest brand names and generic equivalents of common drugs
- Pronunciation provided for all entries
| Source | Date | Version | Tools | Systems named |
|---|---|---|---|---|
| Walkthrough, DEV Community (Sidney Bissoli) | May 12, 2026 | Not stated | 31 | Seven, including optional SNOMED CT |
| Project GitHub repository page | As of October 7, 2026 | Not stated | Not stated | Heading lists six; body describes optional SNOMED CT |
| Official MCP Registry listing | Updated October 6, 2026 | 2.1.1 | 33 | Six in the short description: ICD-11, LOINC, RxNorm, MeSH, ATC, CID-10 |
The registry listing is current metadata for the package. It does not prove how every function behaves at runtime. Check the version you install, and pin it if you need reproducible results, before you quote a tool count or a system count in documentation.
Licensing and deployment checks
- WHO. Live ICD-11 lookup needs WHO credentials. Attribution to WHO is documented in the project repository, and WHO’s own terms apply to the content you retrieve.
- SNOMED CT. The optional tools need a SNOMED International license and a self-hosted Snowstorm backend, because the historical public endpoint has been retired.
- Other terminologies. The repository documents terminology-specific licensing differences. Review the upstream terms for each system you use, especially if you redistribute terminology data or use outputs in regulated workflows.
- Test claims. The walkthrough reports a test suite and daily integration checks. These are project-author claims and have not been independently verified.
- Local policy. Clinical and research organizations should set their own rules for which code systems are permitted and who approves a change to a code.
Where the walkthrough stops
Bissoli’s walkthrough is direct about the tool’s role: “None of this is a substitute for clinical judgment. It’s a lookup layer for already-known codes, not a diagnostic tool.” That framing fits the three examples. Each one moves from a question to a candidate code, class, or subject heading, and each stops short of the decision that depends on it. Used that way, the server saves time on lookups and makes the steps visible; it does not replace the people who make the decision.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →
