EvidenceMD MCP Server for Claude Clinical reasoning, inside Claude

Add EvidenceMD to Claude once, and Claude can ask EvidenceMD whenever you need evidence. You get clinical answers that cite clinical guidelines, clinical journals, and peer-reviewed journals, right inside your Claude chat. It works like a plug-in, and setup takes about a minute.

Updated

MCP server URL
https://evidencemd.ai/mcp
Connect in Claude How it works

Shortened excerpt of a real EvidenceMD answer returned to an MCP client.

Quick Answer

To use EvidenceMD in Claude, add the EvidenceMD MCP server as a custom connector: in Claude open Customize → Connectors → Add custom connector, enter https://evidencemd.ai/mcp, and sign in with your EvidenceMD account. Claude then answers clinical questions with EvidenceMD's clinical reasoning, citing clinical guidelines, clinical journals, and peer-reviewed journals. It works on every Claude plan, including Free, and each question counts toward your EvidenceMD plan. The same server also works in Claude Code, Codex, and Cursor.

Connect EvidenceMD to Claude in about a minute

You need a Claude account and an EvidenceMD account. Add the connector from claude.ai or Claude Desktop, and it follows you to the mobile apps.

Claude Free, Pro, and Max

Add it yourself

  1. 1In Claude, open Customize, then Connectors.
  2. 2Click + and choose Add custom connector.
  3. 3Name it EvidenceMD and paste https://evidencemd.ai/mcp, then click Add.
  4. 4Click Connect and sign in with your EvidenceMD account (Google or email and password).
  5. 5In any chat, turn EvidenceMD on from the + menu and ask a clinical question.

Claude's Free plan allows one custom connector. Pro and Max allow as many as you like.

Claude Team and Enterprise

An owner adds it once

Owner

  1. 1Open Organization settings, then Connectors.
  2. 2Click Add, choose Custom, then Web.
  3. 3Enter https://evidencemd.ai/mcp and click Add.

Each clinician

  1. 1Find EvidenceMD under Customize, then Connectors, and click Connect.
  2. 2Sign in with your own EvidenceMD account and start asking.

Signed up to EvidenceMD with Google? Choose Continue with Google when Claude opens the sign-in page. Signed up with email? Use the same email and password. No EvidenceMD account yet? Create one free.

Use EvidenceMD in Claude Code, Codex, and Cursor

The same MCP server works in the coding tools and agents you already use. Add it once, sign in with your EvidenceMD account, and ask clinical questions from your terminal or editor. Questions count toward your plan exactly as they do in Claude.

Terminal and IDE

Claude Code

  1. 1Run the command once. It adds EvidenceMD for all your projects with a 180-second timeout.
  2. 2In Claude Code, run /mcp, choose evidencemd, and sign in with your EvidenceMD account.

Already added EvidenceMD on claude.ai? It appears in Claude Code automatically when you log in with the same Claude account. Start Claude Code with MCP_TOOL_TIMEOUT=180000 so long answers have time to finish; Claude Code waits 60 seconds by default.

Terminal
claude mcp add-json --scope user evidencemd '{"type":"http","url":"https://evidencemd.ai/mcp","timeout":180000}'
Codex CLI, IDE extension, and ChatGPT desktop app

Codex

  1. 1Add the entry to ~/.codex/config.toml.
  2. 2Run codex mcp login evidencemd and sign in with your EvidenceMD account.

The Codex CLI, IDE extension, and ChatGPT desktop app share this file, so one setup covers all three. You can also add it under Settings, MCP servers, Add server (Streamable HTTP), then click Authenticate. Keep tool_timeout_sec = 180; Codex waits 60 seconds by default.

~/.codex/config.toml
[mcp_servers.evidencemd]
url = "https://evidencemd.ai/mcp"
tool_timeout_sec = 180
Terminal
codex mcp login evidencemd
Editor and agent

Cursor

  1. 1Add the entry to ~/.cursor/mcp.json, or use Cursor Settings, MCP, Add server.
  2. 2Click Connect next to evidencemd and sign in with your EvidenceMD account.

No timeout setting is needed in the Cursor editor.

~/.cursor/mcp.json
{
  "mcpServers": {
    "evidencemd": {
      "url": "https://evidencemd.ai/mcp"
    }
  }
}

What happens when you ask

Claude stays your conversation partner. EvidenceMD does the clinical reasoning and brings the evidence.

  1. 1

    You ask Claude

    Describe the case or question in your own words, in any Claude chat.

  2. 2

    Claude calls EvidenceMD

    Claude sends a complete clinical question to the clinical_reasoning tool.

  3. 3

    EvidenceMD reasons over evidence

    The Expert engine searches clinical guidelines, clinical journals, and peer-reviewed journals and builds the answer.

  4. 4

    You get a cited answer

    Claude shows the answer with numbered source links. It is saved to your history too.

Why clinicians add EvidenceMD to Claude

General assistants are fluent. EvidenceMD makes the answer clinical and verifiable.

The same Expert engine as evidencemd.ai

Answers come from EvidenceMD's Expert clinical reasoning, built for differential diagnosis, workup, and treatment decisions rather than general chat.

Every claim linked to its source

Citations arrive as numbered links to clinical guidelines, clinical journals, and peer-reviewed journals, so you can open and verify each one.

Specialty-aware and multilingual

Claude can pass your specialty and preferred language, and EvidenceMD tailors the answer to both.

Your plan and your history

Questions use your EvidenceMD plan and appear in your EvidenceMD history as their own chats, ready to revisit on the web or in the app.

Never charged for a failed answer

If an answer fails or runs out of time, the question is released and not counted toward your plan.

Secure sign-in, no API keys

Connect with OAuth 2.1 using the EvidenceMD account you already have. Claude never sees your password.

Uses the EvidenceMD plan you already have

Connecting is free. Each question Claude sends to EvidenceMD counts as one clinical reasoning question, exactly like asking on evidencemd.ai.

Free

A limited number of clinical reasoning questions to try it out.

Individual

50 clinical reasoning questions per month, shared between the web, the app, and Claude.

Max

Unlimited clinical reasoning questions, in Claude and everywhere else.

Answers that fail or run out of time are never counted. Compare plans

Privacy and security

  • Secure OAuth 2.1 sign-in: you sign in on evidencemd.ai and Claude never sees your password. Disconnect EvidenceMD in Claude at any time to remove its access.
  • EvidenceMD receives only the question Claude sends to the tool, not the rest of your Claude conversation.
  • Questions and answers are stored encrypted at rest in your EvidenceMD history.
  • What you type into Claude is governed by your Claude plan and Anthropic's terms, so follow your organization's policy before sharing identifiable patient information.

EvidenceMD supports clinical decision-making by qualified professionals and does not replace clinical judgment.

Frequently asked questions

What is the EvidenceMD MCP server?

The EvidenceMD MCP server is the official Model Context Protocol (MCP) connector for EvidenceMD. It brings EvidenceMD's evidence-based clinical reasoning into Claude, Claude Code, Codex, and Cursor: the assistant sends your clinical question to EvidenceMD's clinical_reasoning tool, and EvidenceMD returns a complete answer with inline citations to clinical guidelines, clinical journals, and peer-reviewed journals. The server URL is https://evidencemd.ai/mcp.

How do I add EvidenceMD to Claude?

In Claude, open Customize, then Connectors, click the + button and choose Add custom connector. Enter the name EvidenceMD and the URL https://evidencemd.ai/mcp, click Add, then Connect and sign in with your EvidenceMD account using Google or your email and password. In any chat, turn EvidenceMD on from the + menu and ask a clinical question. On Claude Team and Enterprise plans, an owner adds the connector once under Organization settings, then Connectors, and each clinician connects their own account.

Which Claude plans support the EvidenceMD connector?

Custom connectors work on every Claude plan: Free, Pro, Max, Team, and Enterprise. Claude's Free plan allows one custom connector. The connector works in claude.ai, Claude Desktop, and the Claude mobile apps; add it from the web or desktop app first and it is available everywhere you use Claude.

Is the EvidenceMD Claude connector free?

Connecting is free. Each question Claude sends to EvidenceMD counts as one clinical reasoning question on your EvidenceMD plan, exactly like asking on evidencemd.ai: the Free plan includes a limited number of questions, the Individual plan includes 50 clinical reasoning questions per month, and the Max plan is unlimited. A question that fails or runs out of time is not counted.

What does the EvidenceMD connector return to Claude?

A complete, evidence-based clinical answer from the same Expert clinical reasoning engine that powers evidencemd.ai, with every citation written as a numbered link to its source, such as a clinical guideline, a clinical journal, or a peer-reviewed journal article. Claude can show, quote, and cite that evidence. Each question and answer is also saved to your EvidenceMD history as its own chat.

Does EvidenceMD in Claude handle follow-up questions?

Yes. Claude keeps the conversation, and each EvidenceMD question is answered on its own. For a follow-up, Claude either answers from the EvidenceMD answer it already has or sends EvidenceMD a new, complete question that includes the patient details the follow-up depends on.

Does EvidenceMD work in Claude Code, Codex, and Cursor?

Yes. The same MCP server at https://evidencemd.ai/mcp works in Claude Code, Codex (the Codex CLI, the Codex IDE extension, and the ChatGPT desktop app), and Cursor. In Claude Code, add it with claude mcp add-json and a 180-second timeout, then run /mcp to sign in. In Codex, add an [mcp_servers.evidencemd] entry with tool_timeout_sec = 180 to ~/.codex/config.toml, then run codex mcp login evidencemd. In Cursor, add it to ~/.cursor/mcp.json and click Connect. Each tool signs in with your EvidenceMD account, and the longer timeout matters because Claude Code and Codex stop waiting after 60 seconds by default while detailed answers can take longer.

Is the EvidenceMD MCP connector secure?

You sign in with OAuth 2.1 and PKCE using your existing EvidenceMD account, so Claude never sees your password. EvidenceMD receives only the question Claude sends to the tool, stores questions and answers encrypted at rest in your history, and applies your plan's limits. What you type into Claude itself is governed by your Claude plan and Anthropic's terms, so follow your organization's policies before sharing identifiable patient information.

What is MCP (Model Context Protocol)?

MCP is an open standard that lets AI assistants such as Claude call external tools securely. An MCP server publishes tools, the assistant decides when to use them, and you see the result in the conversation. The EvidenceMD MCP server publishes one tool, clinical_reasoning, which answers clinical questions with evidence and citations.

Documentation and sources

Anthropic: custom connectors using remote MCPClaude's official guide to adding a custom connector on Free, Pro, Max, Team, and Enterprise plans.Model Context ProtocolThe open standard that lets AI assistants call external tools such as EvidenceMD.EvidenceMD pricingFree, Individual, and Max plans. Questions asked in Claude count exactly like questions on evidencemd.ai.EvidenceMD Trust CenterSecurity, privacy, and compliance details for the EvidenceMD platform.EvidenceMD Medical AI APIBuilding your own product? The OpenAI-compatible API returns the same cited clinical reasoning.

Bring EvidenceMD into Claude

Clinical reasoning with citations, right where you already work. Free to connect.

Connect in Claude Create a free account
MCP server URL
https://evidencemd.ai/mcp