openfda-intelligence-mcp

Every FDA drug recall, adverse event, and safety action — structured for AI.

---

What Is This?

The FDA's openFDA program contains over 20 million records across drug adverse events, recalls, product labels, medical device malfunction reports, and food enforcement actions. This data drives billion-dollar regulatory decisions — but the raw API is complex, inconsistently structured, and completely inaccessible to AI workflows.

openfda-intelligence-mcp wraps all five major openFDA endpoints into a single, consistent, MCP-compatible actor. One tool call gives your AI assistant access to the same underlying data that powers FDA safety surveillance, pharmacovigilance teams, and medtech regulatory filings.

Whether you're a pharma compliance officer running weekly signal detection, a medtech startup doing pre-submission due diligence, or a healthcare journalist investigating a drug's safety record — this actor gives you a structured, AI-queryable interface to the FDA's entire public safety database.

---

The Five Endpoints

Endpoint	FDA Source	Records Available	What It Contains
Adverse Events (FAERS)	FDA Adverse Event Reporting System	18M+ reports	Patient-reported drug side effects, outcomes, seriousness
Drug Recalls (Enforcement)	FDA Enforcement Reports	50,000+ actions	Recall class, reason, product description, firm name
Drug Labels	DailyMed / SPL	140,000+ labels	Full prescribing information, contraindications, warnings
Device Events (MDR)	Medical Device Reporting	9M+ reports	Device malfunctions, injuries, deaths linked to medical devices
Food Enforcement	FDA Enforcement Reports	15,000+ actions	Food recalls, market withdrawals, safety alerts

---

Who Uses This?

User Type	Specific Use Case	Value Delivered
Pharma Compliance Teams	Weekly FAERS signal detection for marketed drugs	Catch safety signals 2–3 weeks before formal FDA safety communications
Medtech Startups	Pre-submission 510(k) competitive device safety analysis	Pull MDR reports for predicate devices to anticipate FDA questions
Healthcare AI Builders	Training data for clinical decision support	Structured adverse event data with MedDRA coding
Investigative Health Journalists	Drug safety investigative reporting	Cross-reference FAERS reports with enforcement actions
Biotech Due Diligence Analysts	Safety profile of acquired drug candidates	Historical adverse event burden vs. comparator drugs
Hospital Pharmacy Teams	Drug recall monitoring	Real-time recall alerts for formulary drugs
Regulatory Affairs Consultants	Label comparison and gap analysis	Pull competitor labels side-by-side
Medical Malpractice Attorneys	Adverse event history for litigation	Document FDA-received adverse event frequency

---

Feature Overview

Feature	Detail
Adverse Event Search	Filter by drug name, reaction, seriousness, report date, outcome
Recall Search	Filter by product name, recall class (I/II/III), firm, date
Label Search	Full-text search across 140K+ SPL drug labels
Device Event Search	Filter by device name, manufacturer, event type, date
Food Enforcement Search	Filter by product type, recalling firm, reason for recall
MedDRA Code Filtering	Filter adverse events by standardized MedDRA reaction codes
Seriousness Flags	Filter by serious/non-serious, death, hospitalization, life-threatening
Date Range Queries	All endpoints support receive date and action date filtering
Aggregate Statistics	Count queries: top reactions for a drug, top recalled products by firm
Batch Mode	Run bulk queries across multiple drugs/devices
MCP Real-Time	Live tool calls from Claude Desktop and compatible AI clients

---

Quick Start

MCP Mode

{
  "mcpServers": {
    "openfda-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "@apify/mcp-client",
        "--actor-id",
        "your-username/openfda-intelligence-mcp",
        "--token",
        "YOUR_APIFY_TOKEN"
      ]
    }
  }
}

Ask Claude:

"Search FDA adverse events for metformin. Focus on serious events involving lactic acidosis reported in the last 3 years. Show me the count trend year-over-year."

Batch Mode — Drug Safety Monitor

{
  "mode": "batch",
  "formularyDrugs": ["metformin", "semaglutide", "apixaban", "rosuvastatin"],
  "checkEndpoints": ["adverseEvents", "recalls"],
  "dateFrom": "2024-01-01",
  "seriousOnly": true
}

API Mode

curl -X POST https://api.apify.com/v2/acts/your-username~openfda-intelligence-mcp/run-sync \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "adverseEvents",
    "drugName": "ozempic",
    "reactionTerm": "pancreatitis",
    "seriousOnly": true,
    "dateFrom": "2021-01-01",
    "maxResults": 100
  }'

---

MCP Tools Reference

search_adverse_events

Search the FDA Adverse Event Reporting System (FAERS).

Parameters:

Parameter	Type	Required	Description
drugName	string	No	Generic or brand drug name
reactionTerm	string	No	Adverse reaction term (MedDRA preferred term)
seriousOnly	boolean	No	Filter to serious events only (default: false)
outcomes	array	No	Filter by outcome: death, hospitalization, life_threatening, disability, congenital_anomaly
reportDateFrom	string	No	Report receive date start (YYYYMMDD)
reportDateTo	string	No	Report receive date end (YYYYMMDD)
ageGroup	string	No	pediatric, adult, elderly
sex	string	No	male, female
maxResults	integer	No	Max results (default: 50, max: 1000)
countBy	string	No	Return aggregate count by: reaction, drug, year, outcome

Example response:

{
  "adverseEvents": [
    {
      "safetyReportId": "US-PFIZER-2023-1234567",
      "receiveDate": "20230615",
      "serious": true,
      "seriousnessHospitalization": true,
      "primaryDrug": {
        "name": "OZEMPIC",
        "genericName": "semaglutide",
        "indication": "TYPE 2 DIABETES",
        "dosage": "0.5 MG/DOSE",
        "route": "SUBCUTANEOUS"
      },
      "reactions": [
        { "term": "Pancreatitis acute", "outcome": "Recovered/resolved" }
      ],
      "patientAge": 58,
      "patientSex": "Female",
      "country": "US",
      "reporterQualification": "Physician"
    }
  ],
  "totalCount": 2847,
  "queryMetadata": { "dataSource": "openFDA FAERS", "coverage": "1990-present" }
}

search_drug_recalls

Search FDA drug enforcement and recall actions.

Parameters:

Parameter	Type	Required	Description
productDescription	string	No	Product name or description search
recallingFirm	string	No	Company issuing the recall
recallClass	string	No	I (dangerous), II (may harm), III (unlikely to harm)
reasonForRecall	string	No	Text search in reason field
initiationDateFrom	string	No	Recall initiation date start
initiationDateTo	string	No	Recall initiation date end
voluntaryMandated	string	No	Voluntary or FDA Mandated
maxResults	integer	No	Max results

search_drug_labels

Search FDA-approved drug prescribing information.

Parameters:

Parameter	Type	Required	Description
brandName	string	No	Drug brand name
genericName	string	No	Generic drug name
activeIngredient	string	No	Active ingredient search
manufacturer	string	No	Manufacturer name
labelSection	string	No	Specific section: warnings, contraindications, adverse_reactions, dosage, indications
searchText	string	No	Full-text search across entire label

search_device_events

Search FDA Medical Device Reports (MDRs).

Parameters:

Parameter	Type	Required	Description
deviceName	string	No	Device brand or generic name
manufacturer	string	No	Device manufacturer name
eventType	string	No	malfunction, injury, death, other
deviceClass	string	No	FDA device class: I, II, III
productCode	string	No	FDA product code
dateFrom	string	No	Event date start
dateTo	string	No	Event date end
maxResults	integer	No	Max results

search_food_enforcement

Search FDA food recall and enforcement actions.

Parameters:

Parameter	Type	Required	Description
productDescription	string	No	Food product name or description
recallingFirm	string	No	Company issuing the recall
reasonForRecall	string	No	Reason text search (e.g., "Listeria", "undeclared allergen")
recallClass	string	No	I, II, or III
dateFrom	string	No	Recall initiation date start
dateTo	string	No	Recall initiation date end

---

Use Case Recipes

Recipe 1: Pharma Compliance — Weekly Signal Detection

Scenario: Your drug Luxivatrex was approved 18 months ago. You need to monitor FAERS weekly for new serious adverse event signals, especially hepatic events.

Prompt:

"Search FDA adverse events for the drug 'luxivatrex' from the past 7 days. Focus on serious events. Categorize by reaction term, show me counts, and flag any liver-related terms (hepatic failure, elevated transaminases, jaundice) immediately."

Batch config for weekly monitoring:

{
  "mode": "batch",
  "monitorDrugs": ["luxivatrex", "competitor-drug-a"],
  "endpoints": ["adverseEvents"],
  "seriousOnly": true,
  "lookbackDays": 7,
  "alertKeywords": ["hepatic failure", "liver", "jaundice", "transaminase"]
}

Recipe 2: Medtech 510(k) Pre-Submission Due Diligence

Scenario: You're submitting a 510(k) for a new insulin infusion pump. Your predicate device is the MedTech Pro 3000. You need to characterize its MDR history.

Prompt:

"Pull all FDA Medical Device Reports for the 'MedTech Pro 3000' insulin infusion pump from the past 5 years. Categorize by event type (malfunction vs injury vs death). What are the most common malfunction modes? What percentage resulted in patient injury?"

Recipe 3: Investigative Journalism — Drug Safety Pattern

Prompt:

"I'm investigating safety issues with GLP-1 drugs. Search adverse events for semaglutide AND tirzepatide AND liraglutide from 2021 to present. For each: total serious events, top 5 reactions, death count. Then check if any of these drugs have active FDA recalls. Give me a comparison table."

Recipe 4: Hospital Pharmacy — Formulary Recall Monitor

Prompt:

"Check for any FDA Class I or Class II drug recalls initiated in the last 30 days for these formulary drugs: metoprolol, lisinopril, atorvastatin, metformin, amlodipine, omeprazole, levothyroxine. If any recalls found, give me the lot numbers and recall reason."

Recipe 5: Biotech M&A Due Diligence

Prompt:

"We're acquiring a company that markets the drug 'Xorambix'. Pull everything: full adverse event history, any recalls, the current approved label, and flag any FDA warning letters visible in the enforcement database. Summarize the safety burden I'm inheriting."

---

Claude Desktop Configuration

{
  "mcpServers": {
    "openfda-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "@apify/mcp-client",
        "--actor-id",
        "your-username/openfda-intelligence-mcp",
        "--token",
        "apify_api_XXXXXXXXXXXXXXXXXXXX"
      ]
    }
  }
}

Test query after setup: "Search FDA adverse events for aspirin, top 10 reactions by count."

---

Pricing & Usage

Tier	Price	Includes
Free	$0	200 API calls/month across all endpoints
Starter	$19/month	5,000 calls/month
Professional	$79/month	30,000 calls/month + batch mode
Compliance	$299/month	Unlimited + weekly monitoring + email alerts

---

Data Source & Freshness

• FAERS: Updated quarterly; latest quarter typically available within 90 days of close
• Enforcement (Recalls): Near real-time; updated weekly
• Drug Labels: Updated as SPL submissions are processed; typically within days of FDA approval
• MDR (Device Events): Approximately 2–3 week lag from event submission
• Food Enforcement: Updated weekly
• Total dataset coverage: Adverse events from 1990; enforcement from 2004; labels from 2009

---

Technical Notes

• All data sourced directly from api.fda.gov — no third-party intermediaries
• openFDA enforces 1,000 results per query limit; this actor handles pagination automatically
• FAERS data represents reported events, not confirmed causal relationships — appropriate disclaimers are embedded in all responses
• Drug name matching includes both brand and generic name normalization
• MedDRA coding (where present in FAERS) is preserved in structured output
• Device event "event type" field is normalized across FDA's inconsistent historical coding

---

Support & Issues

• GitHub: github.com/your-username/openfda-intelligence-mcp
• Documentation: Inline with MCP tool schemas
• Email: support@your-domain.com

Data sourced from openFDA (api.fda.gov). Not affiliated with the US Food and Drug Administration. Adverse event data does not constitute evidence of causality.