RegDataLab – FDA Compliance Data

Check any supplier or company against public FDA records: inspections, Form 483s, warning letters, recalls and import alerts across the whole company family, each linked to fda.gov.

Lihat di regdatalab.com

Server MCP Terhosting

npx add-mcp 'https://www.regdatalab.com/mcp'

Terpasang ke Claude Code, Codex, Cursor, dan lainnya

Dokumentasi

RegDataLab MCP server (fda-data-mcp)

Public FDA compliance records (inspections, Form 483s, warning letters, recalls) linked by company and site, for Claude, ChatGPT, any MCP client or API.

This is the public install and discovery repo for RegDataLab – FDA Compliance Data, the hosted MCP server at https://www.regdatalab.com/mcp. It has 51 read-only tools over 35 public FDA datasets, and every record links back to its FDA source. The npm package keeps the name fda-data-mcp. The private backend, ingestion system, and production data pipeline are not open-sourced here.

One question, the whole company family

Ask about a supplier by the name you know. RegDataLab finds the subsidiaries FDA files under other names, then answers across all of them, with a link to the FDA source on every record.

You ask: "Check Baxter as a supplier."

One call to fda_manufacturing_risk_summary searches Baxter and 21 linked companies, including Hillrom and Welch Allyn:

	Across the Baxter family
Sites	65 FDA-registered or inspected sites
Inspections	201 on record: 103 NAI, 72 VAI, 26 OAI. Latest OAI: Round Lake, IL, April 2026
Warning letters	15, most recent September 29, 2026
Form 483s	Latest published: Baxter Healthcare, Round Lake and Baxter Oncology GmbH
Recalls	703 distinct recalls, 324 in the last 5 years
Import alerts	1 red-list entry on Import Alert 66-40 (2014)

From a real answer on November 12, 2026. Searching FDA's own sites for "Baxter" would miss the Hillrom and Welch Allyn records. NAI, VAI and OAI are FDA's inspection outcomes: no action, voluntary action, official action indicated.

Try it now, no signup

Paste this into a terminal with Claude Code. It uses the shared demo key from the regdatalab.com home page:

claude mcp add regdatalab-demo https://www.regdatalab.com/mcp \
  --transport http \
  --header "Authorization: Bearer fda_demo_GKNDjUOjPbhLZZVPvRHDlY3zP_H3HXzTMRbHVD_I5rg"

Then ask: "Check Baxter as a supplier." The demo key is shared by every visitor and capped at 20 requests per minute and 50 credits per IP address. For your own key, sign up free (50 credits, no card), or add the bare URL and sign in as shown under Quick Start.

The same question in ChatGPT, from question to sourced answer:

A question about Pfizer answered in ChatGPT with RegDataLab: inspections, Form 483s, recalls and warning letters, each with its source

Full walkthrough (4 minutes): regdatalab-chatgpt-review-walkthrough.mp4. No terminal? Look up any company at regdatalab.com.

RegDataLab homepage screenshot

Quick Links

Website: regdatalab.com
MCP endpoint: https://www.regdatalab.com/mcp
Discovery metadata: https://www.regdatalab.com/.well-known/mcp.json
Connect guide: regdatalab.com/connect.md
Product docs: regdatalab.com/docs
Pricing: regdatalab.com/pricing
Signup: regdatalab.com/signup
Changelog: CHANGELOG.md
Current tool list: regdatalab.com/api
Postman workspace with ready-made calls: RegDataLab FDA Compliance Data
Free CSV dataset (FDA warning letters and OAI inspections, 2021-2026, CC0): Hugging Face and Kaggle
Wrapper package source: package.json

What It Does

RegDataLab gives AI agents structured access to FDA data for questions like:

Which facilities does this company operate?
Has FDA inspected them recently?
Were those inspections NAI, VAI, or OAI, and what did the Form 483s cite?
Are there warning letters, recalls, import alerts, import refusals or debarments?
What does FDA show for 510(k), PMA, Drugs@FDA, NDC, and related regulatory records?

The strongest current use case is manufacturing and compliance intelligence for pharma, biotech, and medtech teams.

Quick Start

The hosted endpoint is https://www.regdatalab.com/mcp. Pick the first path your client supports:

OAuth sign-in (claude.ai, Claude Desktop, Claude Code, ChatGPT, Cursor, VS Code, Codex and other OAuth-capable clients): add the bare URL and sign in. No key in your config.
Authorization header (CI, scripts, the OpenAI API, clients without sign-in): send Authorization: Bearer YOUR_API_KEY.
npx wrapper (stdio-only clients): set FDA_DATA_API_KEY.

Never put an API key in a URL: the server refuses a key in the query string. The canonical, current client steps are in the connect guide.

claude.ai, Claude Desktop and Cowork

RegDataLab is listed in Claude's connector directory.

Open Customize > Connectors and search for RegDataLab under Discover, or choose Add custom connector and enter only https://www.regdatalab.com/mcp. On Team and Enterprise plans an owner adds it in Organization settings > Connectors, then members click Connect.
On the RegDataLab consent page, paste your RegDataLab API key and approve read-only access.
Turn it on in a chat under + > Connectors.

ChatGPT

RegDataLab is not in ChatGPT's app directory yet, so add it as a custom MCP server: on chatgpt.com open Plugins, click Add, choose Add custom MCP server, enter only https://www.regdatalab.com/mcp with Authentication set to OAuth, and sign in. Your plan or workspace must allow custom MCP servers. Then type @RegDataLab in a chat.

Claude Code

claude mcp add --transport http --scope user regdatalab https://www.regdatalab.com/mcp

Run /mcp, pick regdatalab and choose Authenticate. For CI or a machine with no browser, send the key as a header instead:

claude mcp add regdatalab https://www.regdatalab.com/mcp \
  --transport http \
  --header "Authorization: Bearer YOUR_API_KEY" \
  --scope user

Cursor

Add to ~/.cursor/mcp.json, then open Settings > MCP and click Connect:

{
  "mcpServers": {
    "regdatalab": {
      "url": "https://www.regdatalab.com/mcp"
    }
  }
}

To use an API key instead of signing in (Cursor expands ${env:VAR}):

{
  "mcpServers": {
    "regdatalab": {
      "url": "https://www.regdatalab.com/mcp",
      "headers": { "Authorization": "Bearer ${env:REGDATALAB_API_KEY}" }
    }
  }
}

VS Code

Or add to .vscode/mcp.json or your user mcp.json (the key is servers):

{
  "servers": {
    "regdatalab": {
      "type": "http",
      "url": "https://www.regdatalab.com/mcp"
    }
  }
}

Codex CLI, Codex IDE extension and ChatGPT desktop

codex mcp add regdatalab --url https://www.regdatalab.com/mcp
codex mcp login regdatalab

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "regdatalab": {
      "serverUrl": "https://www.regdatalab.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

OpenAI API (Responses)

from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model="gpt-4.1",
    tools=[{
        "type": "mcp",
        "server_label": "regdatalab",
        "server_url": "https://www.regdatalab.com/mcp",
        "require_approval": "never",  # tools are read-only
        "headers": {"Authorization": "Bearer YOUR_API_KEY"},
    }],
    input="Give me a manufacturing risk summary for Pfizer",
)
print(response.output_text)

Generic MCP clients

URL: https://www.regdatalab.com/mcp (Streamable HTTP)
Auth: OAuth 2.1 with PKCE S256 and dynamic client registration (scope fda:read), or the header Authorization: Bearer YOUR_API_KEY (or x-api-key)
Persist the Mcp-Session-Id returned by initialize
Discovery file: https://www.regdatalab.com/.well-known/mcp.json

npx wrapper (stdio-only clients)

For clients that can only start a local command (needs Node.js 18 or later). In Claude Desktop, the connector you add in claude.ai already appears with nothing to install; use this only for a local stdio setup, such as claude_desktop_config.json:

{
  "mcpServers": {
    "regdatalab": {
      "command": "npx",
      "args": ["-y", "fda-data-mcp"],
      "env": {
        "FDA_DATA_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Config file location:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Or from a shell: FDA_DATA_API_KEY=YOUR_API_KEY npx -y fda-data-mcp

The wrapper is a zero-dependency stdio-to-HTTP proxy:

source: bin/fda-data-mcp.js
proxy: lib/stdio-proxy.js

The version on npm can lag this repo; see CHANGELOG.md for what each published version contains.

If you do not have an API key yet, sign up at regdatalab.com/signup. A new key starts with 50 credits; verifying your email unlocks 300 free credits a month.

Example Prompts

Give me a manufacturing risk summary for Pfizer.
Show recent VAI and OAI inspections for Moderna.
Summarize recalls, compliance actions, and import refusals for Thermo Fisher.
What does FDA have on this company across facilities, inspections, and enforcement?

Data Coverage

RegDataLab reads 35 public FDA datasets, including:

FDA inspections and inspection classifications (NAI, VAI, OAI)
Form 483 citations by CFR section
warning letters, untitled letters and other compliance actions
recalls and enforcement reports
import alerts and import refusals
debarments
facility registrations and product listings (drugs and devices)
510(k) clearances, De Novo grants and PMA approvals
Drugs@FDA, NDC directory, NSDE and drug labels
device classification and UDI records
drug and device shortages
adverse event reports
MES, QMS and LIMS software vendor evidence (software_* tools)

Company lookups match legal names, former names and common short names (for example "Philips"), and can widen to a company's corporate family.

For the current tool list and dataset freshness, use:

API page (authoritative tool list)
Docs
Status

Auth and Pricing

Sign up for an API key at regdatalab.com/signup
OAuth-capable clients: add the bare URL and sign in; no key in your config
Other HTTP clients: send the Authorization: Bearer YOUR_API_KEY header
The npx wrapper: set FDA_DATA_API_KEY=YOUR_API_KEY
Never put a key in a URL or in an issue report
Each tool call uses credits; calls that fail on bad input or a server error are free
A new account starts with 50 credits; verifying your email adds 300 free credits a month. No card required
Paid plans are listed at regdatalab.com/pricing

Why This Repo Exists

This repo is intentionally public and lightweight so it can serve as:

the GitHub landing page for the hosted MCP
a stable place for setup examples
the home of the official MCP Registry entry (server.json, see REGISTRY.md) and the fda-data-mcp npm wrapper

The production backend and ingestion engine live separately.

Data source and independence

RegDataLab is an independent data product, not affiliated with, endorsed by, or sponsored by the U.S. Food and Drug Administration. Data provided by the U.S. Food and Drug Administration, through openFDA and other public FDA sources. RegDataLab does not provide medical advice. RegDataLab is operated by Wasatch Crest Holdings LLC.

Support

Changelog and Releases

Version history lives in CHANGELOG.md

Security

If you find a security issue, do not open a public issue. Use the policy in SECURITY.md.