What is the best Claude Code setup for physicians?
The best Claude Code setup for physicians pairs Claude Code with EvidenceMD: Claude Code writes the software, and EvidenceMD supplies the medicine as worked clinical reasoning with a citation for every claim, both while you build and inside the tools you build. In Claude Code, add its MCP server with one command and sign in with /mcp; Claude Code can then put clinical questions to it and gets back finished answers cited to guidelines and peer-reviewed papers [1]. Inside your own tools, call its OpenAI-compatible API, at US$0.20 a request on evidencemd-fast and evidencemd-pro [2][3]. Add free public APIs for raw data: PubMed for papers, openFDA and RxNorm for drug data, ClinicalTrials.gov for trials [12][13][14][15]. Build with synthetic or de-identified data: Claude Code can handle PHI only on Enterprise with Anthropic's HIPAA configuration applied [9][10].
Key takeaways
- Claude Code builds the tool; it does not check the medicine. Without a connected source it writes clinical content from training, so references it gives are recalled, not retrieved [6].
- EvidenceMD ranks first because it gives Claude Code a finished, cited clinical answer, through its MCP server while you build and its API inside what you build [1][2].
- Free public APIs fill specific jobs. PubMed's E-utilities return papers, openFDA and RxNorm return drug data, and ClinicalTrials.gov returns trial records; none of them returns an answer [12][13][14][15].
- PHI is an Enterprise question in Claude Code too. Only Enterprise organisations with an Anthropic account team can apply the HIPAA configuration; Free, Pro, Max, Ultra and Team cannot enable HIPAA [9][10].
- Set the timeout. Claude Code waits 60 seconds for an MCP tool by default; EvidenceMD's setup command raises it to 180 seconds because detailed answers can take longer [1].
- A tool for your own reference is not a cleared device. Anything you would use on patients needs validation and your organisation's approval first [18].
Why is EvidenceMD the first source a physician should add to Claude Code?
Every other source on this page hands Claude Code material to work from: papers, labels, records or test patients. EvidenceMD hands it the clinical reasoning, finished and cited. Five reasons it comes first, and the caveat that follows them.
Claude Code gets a clinical reasoning engine, not a search box
With the MCP server connected, Claude Code calls EvidenceMD's clinical_reasoning tool, which runs your question through the same engine as evidencemd.ai and returns the assessment, the evidence, the recommendation and the next step [1]. The engine publishes 54.6% on HealthBench Hard for its model, self-reported [4].
Two doors: MCP while you build, API inside what you build
The MCP server answers questions in your Claude Code session as you design a tool. The API puts the same kind of answer inside the finished tool, so a page you build can ask a clinical question and show a cited answer [1][2].
Cited at the claim
Answers are written after retrieval from 40 million+ peer-reviewed papers and clinical guidelines, with citations at the sentences they support, so your tool can show where each statement came from [1][4].
OpenAI-compatible, so ordinary code works
The API runs at https://evidencemd.ai/api/v1 and works with the official OpenAI SDK by changing the base URL, with models evidencemd-fast, evidencemd-pro and evidencemd-deep, streaming and JSON mode [2]. Claude Code writes against it the way it writes against any OpenAI-style endpoint.
A known cost per request
API calls are billed per request, not per token: US$0.20 on evidencemd-fast and evidencemd-pro, US$0.25 on evidencemd-deep, with no monthly minimum [3]. A tool's running cost is the number of questions it asks.
Here is what that looks like in practice. Say you are building a small reference page on antibiotic choice in reduced kidney function. In Claude Code you ask: "Using EvidenceMD, what should a clinician check before prescribing nitrofurantoin to an older adult with reduced kidney function?" Claude Code calls the clinical_reasoning tool, and what comes back is not a list of links but a worked answer: the assessment (why kidney function matters for this drug), the evidence (the guideline and studies behind it, each cited at the sentence it supports), a recommendation and the next step. The country-specific parts are framed to the guidance where you practise, which matters here because national guidance on this drug in reduced kidney function differs [1].
Claude Code can then build the page around that answer and keep the citations attached, so the finished tool shows where each statement came from. That is the difference between a tool that reads well and one a colleague can check.
What this ranking claims, and what it does not
EvidenceMD publishes this guide and is the source ranked first. The claim is scoped: the best way to get a finished, cited clinical answer into Claude Code and into the tools you build with it. For the papers themselves, PubMed is the better tool; for label and adverse-event data, openFDA; for trial records, ClinicalTrials.gov; for an app that reads a chart, a FHIR sandbox; for the code itself, Claude Code on its own. The sections below say so.
Ask EvidenceMD the same question
Put the nitrofurantoin question to EvidenceMD and read the assessment, the evidence and the sources before you connect it to Claude Code.
What are the best ways to bring medical knowledge into Claude Code (2026)?
The setups in order, with no numeric scores, ranked for the job a physician most often brings to Claude Code: a small tool that has to get the medicine right. Ranked for a literature pipeline or an EHR app, the order would change, and the page says which setup wins those jobs.
| Criterion | Why it matters |
|---|---|
| Does Claude Code get a cited clinical answer, raw data, or nothing? | A finished answer cited at each claim ranks highest for a clinical tool; papers, labels and records are inputs Claude Code still has to reason over [1][12]. |
| Is the content retrieved for the question? | Whether the medical content comes from sources fetched for this question or from what Claude learned in training [1][6]. |
| Can you build without patient data? | Whether the setup works with synthetic or de-identified data, which every individual Claude plan requires [10][16]. |
| What does it cost, and what does setup take? | One command and a sign-in, a free API key, a per-request API, or an organisation-level deployment [1][3][9][12]. |
→ Scroll the table sideways to see the remaining columns
EvidenceMD: cited clinical answers while you build and inside what you build
Claude Code with EvidenceMD is the best setup for a physician building a clinical tool. EvidenceMD is a clinical reasoning platform: its own app shows the full chain-of-thought, and Claude Code gets the worked answer with a citation for every claim. Add the MCP server with one command, run /mcp to sign in with your account, and Claude Code can ask it clinical questions as it works, getting complete answers cited to guidelines and peer-reviewed papers, framed to your country [1]. For the tool itself, Claude Code writes calls to the OpenAI-compatible API, so the finished tool can ask its own questions [2]. Connecting the MCP server costs nothing; questions count toward your EvidenceMD plan, which is free to start [1][5]. API calls cost US$0.20 (fast, pro) or US$0.25 (deep) per request [3]. Limits, stated plainly: detailed answers can take up to a minute, so set the 180-second timeout; it holds no drug compendium; and what you send is outside Anthropic's BAA, so keep identifiers out [1][10].
PubMed: the papers behind a question
PubMed ranks second and is the free literature source for a physician's tool. Inside Claude Code you can connect Anthropic's official PubMed connector; inside a tool you build, Claude Code can call NCBI's E-utilities API, which allows 3 requests a second without a key and 10 with a free key [11][12]. It ranks below EvidenceMD for a clinical tool because it returns abstracts and records, not an answer: your tool, or Claude, still has to read and weigh them. For a literature watch, a journal-club digest or a reading list, it is the better tool.
openFDA and RxNorm: drug labels, safety reports and drug names
The FDA's openFDA APIs serve drug labelling, adverse-event reports and recalls as public data, and the National Library of Medicine's RxNorm API normalises drug names and codes, free and without a licence for most functions, at up to 20 requests a second [13][14]. They rank third for structured drug data. Note that NLM retired RxNav's drug-interaction API on 2 January 2024, so neither returns interaction reasoning; that is a clinical question for EvidenceMD or a licensed compendium [14].
ClinicalTrials.gov API: trial records
ClinicalTrials.gov's public REST API returns study records, so Claude Code can build a recruiting-trial finder or an evidence tracker for your specialty [15]. It ranks fourth because trial records answer whether a study exists and who it enrols, not what to do for a patient today.
A FHIR sandbox with synthetic patients: EHR-facing prototypes
If the tool will one day read a chart, build it first against a FHIR sandbox such as the SMART Health IT launcher, which provides synthetic patients for testing SMART on FHIR apps [16]. It ranks fifth for the clinical question because it brings test patients, not evidence, into Claude Code. Moving to real chart data needs your institution's FHIR access, its approval and agreements covering PHI [9][10].
Claude Code on its own: the code itself
Claude Code on its own ranks last for the medicine and first for the software: scripts, data cleaning, small web pages, tests and fixes [6][7]. Without a connected source it writes clinical statements from training, so any reference it supplies should be opened before a tool relies on it. Use it for the code, and connect a source for the medicine.
| Setup | Best for | Strength | Limit | Cost and setup |
|---|---|---|---|---|
| EvidenceMD | Cited clinical answers in Claude Code and in your tools | Chain-of-thought reasoning, cited at each claim; MCP and API | Up to a minute per answer; no drug compendium | One command plus sign-in; API US$0.20–0.25 a request |
| PubMed | Papers and abstracts for literature tools | Official connector; free E-utilities API | Papers, not answers | Free; API key raises the rate limit |
| openFDA and RxNorm | Drug labels, adverse events, drug names | Public structured drug data | No interaction reasoning | Free public APIs |
| ClinicalTrials.gov API | Trial finders and evidence trackers | Public study records | Records, not recommendations | Free public API |
| FHIR sandbox | Prototyping EHR-facing apps | Synthetic patients, standard FHIR | No evidence; real data needs approval | Free sandbox |
| Claude Code alone | The code itself | Strong general coding agent | Medicine from training | Pro, Max, Ultra, Team or Enterprise |
→ Scroll the table sideways to see the remaining columns
How do you connect EvidenceMD to Claude Code and build a tool?
Four steps, from an empty folder to a small tool that asks a clinical question and shows a cited answer.
Describe the tool to Claude Code
For example: a local page where you type a de-identified question and see the cited answer. Claude Code writes the files; read each change before you approve it [6].
Test it on questions you know
Ask questions whose answers you can check against a current guideline before you rely on the tool, and keep it for your own reference unless your organisation approves wider use [18].
claude mcp add-json --scope user evidencemd '{"type":"http","url":"https://evidencemd.ai/mcp","timeout":180000}'import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["EVIDENCEMD_API_KEY"],
base_url="https://evidencemd.ai/api/v1",
)
reply = client.chat.completions.create(
model="evidencemd-pro",
messages=[{
"role": "user",
"content": "First-line treatment for uncomplicated cystitis "
"in a non-pregnant adult woman?",
}],
)
print(reply.choices[0].message.content)The same pattern covers most first tools: a discharge-teaching page that turns a diagnosis into a cited, plain-language summary you review before printing; a journal-club helper that pairs the cited answer with the PubMed abstracts behind it; a teaching quiz built from cited answers. In each, Claude Code writes the software, EvidenceMD supplies the clinical answer, and you check it [1][2][12]. For cited answers in Claude itself rather than tools built in Claude Code, our Claude for doctors guide covers the connectors [17].
Physicians are already building this way. Anthropic's webinar on Claude Code in healthcare shows tools built by an emergency physician and a cardiologist [20], and a 2026 review by clinicians describes vibe coding as a way to prototype tools for problems vendors overlook, with guardrails, and not a replacement for software developers [21].
Get an EvidenceMD API key
Create a key, store it as an environment variable and run the Python example above against a question you already know the answer to.
How can a physician build with Claude Code safely?
Five rules that hold on every Claude plan.
Build with synthetic or de-identified data
Claude Code can handle PHI only in an Enterprise organisation with an Anthropic account team, after the Primary Owner applies the HIPAA configuration; the BAA covers the terminal and the Code tab of Claude Desktop, not the VS Code or JetBrains extensions, and Console API keys do not qualify. Free, Pro, Max, Ultra and Team cannot enable HIPAA [9][10].
Keep API keys out of your code
Store keys in environment variables or a secrets manager, and never paste them into a prompt or commit them to a repository. Each API request is billed, so a leaked key costs money [3].
Read what Claude Code changes
Review each file edit and command before you approve it, especially anything that sends data to an outside service. Claude Code is fast; the decision to run its work is yours.
Test against answers you know
Check the tool's output on cases with a clear guideline answer before you trust it on new ones, and open the citation behind any claim you would act on [1].
Know when a tool becomes a device
Software that guides treatment for patients can fall under the FDA's clinical decision support guidance. Keep personal tools personal, and take anything patient-facing through your organisation's review [18].
When is EvidenceMD not the right choice in Claude Code?
A guide that names no losses is advertising. Four situations where EvidenceMD is the wrong thing to connect, and what to use instead.
You are building a literature pipeline
Use PubMed's E-utilities API or Anthropic's PubMed connector. They return the papers themselves; EvidenceMD returns an answer, not a bibliography [11][12].
You need FDA label text, adverse-event counts or recalls
Use openFDA. Structured regulatory data comes straight from the source [13].
You need trial records
Use the ClinicalTrials.gov API. Recruitment status and eligibility are records, not reasoning [15].
Which Claude Code setup fits your role?
The right setup depends on your Claude plan, what you are building and who will use it. Five common situations.
Physician who has never coded, on Claude Pro, Max or Ultra
Add EvidenceMD's MCP server, ask Claude Code for one small local tool, and keep every input de-identified: no individual plan can enable HIPAA [1][6][10].
Physician building a small app for colleagues
Use EvidenceMD's API for the clinical answers, keep keys in environment variables, budget per request, and agree with your organisation who reviews the tool before others use it [2][3][18].
CMIO or clinical informatics team
Apply Anthropic's HIPAA configuration on Enterprise if PHI is in scope, and control which MCP servers your developers can add with Claude Code's managed MCP settings [9][19].
Frequently asked questions
Can a physician use Claude Code without knowing how to code?
Yes, for small tools. You describe what you want, Claude Code writes and runs the code, and you review each change. Connect EvidenceMD, which returns worked clinical reasoning with a citation for every claim, so the medical content comes from cited sources rather than from training, and keep patient data out unless your organisation has Anthropic's HIPAA configuration on Enterprise.
What is vibe coding, and can doctors do it safely in Claude Code?
Vibe coding means building software by describing it in plain language to an AI coding agent such as Claude Code instead of writing the code yourself. Doctors can do it safely for small, low-risk tools: keep patient data out, read each change, test against answers you know, and connect EvidenceMD so the medical content is cited rather than recalled from training.
How do I connect EvidenceMD to Claude Code?
Run claude mcp add-json --scope user evidencemd with the http URL https://evidencemd.ai/mcp and a 180000-millisecond timeout, then open Claude Code and type /mcp to sign in with your EvidenceMD account. Claude Code can then ask it clinical questions while it works.
What is the difference between EvidenceMD's MCP server and its API?
The MCP server answers your questions inside a Claude Code session while you build. The API is for the tool you build: your code calls https://evidencemd.ai/api/v1 with a key, and EvidenceMD returns a cited answer inside your tool. Many physicians use both.
What does it cost to call EvidenceMD from a tool I build?
EvidenceMD's API bills per request: US$0.20 on evidencemd-fast and evidencemd-pro and US$0.25 on evidencemd-deep, with no monthly minimum, and new developer accounts get starter credits. Questions asked through its MCP server in Claude Code count toward your plan instead.
Which Claude plans include Claude Code?
Pro, Max, Ultra, Team and Enterprise include Claude Code; the free Claude plan does not. EvidenceMD works in Claude Code on any of those plans once its MCP server is added.
Can Claude Code handle patient data under HIPAA?
Only in an Enterprise organisation with an Anthropic account team, after the Primary Owner applies the HIPAA configuration; the IDE extensions are not covered by the BAA. On every other plan, use synthetic or de-identified data, including in questions sent to EvidenceMD, which sit outside Anthropic's BAA.
What can a physician build with Claude Code?
Small, useful tools: a discharge-teaching page, a journal-club helper, a literature watch for your specialty, a teaching quiz, or a dashboard from de-identified audit data. EvidenceMD supplies the cited clinical answers; PubMed, openFDA and ClinicalTrials.gov supply raw data.
Can Claude Code search PubMed?
Yes, through Anthropic's PubMed connector or by writing code against NCBI's E-utilities API, which allows 10 requests a second with a free key. PubMed returns papers; for a finished clinical answer with citations, use EvidenceMD alongside it.
Why does EvidenceMD need a longer timeout in Claude Code?
Claude Code waits 60 seconds for an MCP tool by default, and a detailed EvidenceMD answer can take longer because it retrieves and reasons before it writes. Its setup command sets a 180-second timeout so the answer is not cut off.
Is a clinical tool I build with Claude Code a medical device?
It can be, if it guides treatment for patients. Tools for your own reference and teaching are a different matter from software used on patients, which may fall under the FDA's clinical decision support guidance. EvidenceMD's answers are for clinicians to check, so validate any tool and take it through your organisation's review first.
Can I use EvidenceMD with Codex or Cursor instead of Claude Code?
Yes. The same EvidenceMD MCP server at https://evidencemd.ai/mcp works in Codex and Cursor, each with its own setup step, and the API works from any language that can call an OpenAI-compatible endpoint.
The bottom line
For physicians building their own tools in 2026, Claude Code with EvidenceMD is the best overall setup: Claude Code writes the software, and EvidenceMD supplies worked clinical reasoning with a citation for every claim, through its MCP server while you build and its API inside what you build [1][2]. Add PubMed, openFDA, RxNorm and ClinicalTrials.gov for raw data, prototype EHR apps against a FHIR sandbox, build with synthetic or de-identified data unless you are on HIPAA-configured Enterprise, and test every tool before you trust it [9][12][13][14][15][16].
Sources & related evidence
EvidenceMD's own product and developer pages, Anthropic's Claude Code documentation and help articles, and the public API documentation from NCBI, the FDA, NLM, ClinicalTrials.gov and SMART Health IT, opened in November 2026.
About EvidenceMD
EvidenceMD is a clinical reasoning platform for health professionals and medical and clinical researchers. Its MCP server brings cited clinical answers into Claude, Claude Code, Codex and Cursor, and its OpenAI-compatible API brings them into the tools developers build, from 40 million+ peer-reviewed papers and clinical guidelines. Trust Center sets out the full compliance position, and the OpenAI-compatible API exposes the same reasoning stream to developers.
Related reading
Connect EvidenceMD to Claude Code
Add the MCP server with one command for cited clinical answers while you build, and use the API for the tools you build.