openFDA MCP. Query Drug, Food, and Device Safety Data.

You're dealing with public health data—the kind of raw records you need for real research, not some polished dashboard view. This MCP Server connects your AI client directly to three massive US government databases through openFDA. You don't get a pre-built API; you send complex Lucene syntax queries and get live results on adverse events, food safety recalls, or medical device malfunctions.

When you use the query_drug_events tool, your agent runs deep searches against millions of reports detailing drug side effects and medication errors. You can investigate specific reactions linked to drugs, whether it's tracking complaints about aspirin-induced headaches across patient records or pinpointing adverse events for a particular class of medicine.

The system processes raw data entries, letting you structure queries around both the reported adverse event and the associated pharmaceutical product.

For food safety, the query_food_recalls tool searches FDA databases for outbreaks and enforcement actions. You can narrow your focus significantly, filtering by specific pathogens like Listeria or searching across entire state records to track current or historical recalls. It lets you check records based on a product type, a specified reason for the recall, or even the status of the investigation itself.

The query_medical_devices tool pulls safety reports—the MAUDE data—on medical devices. You'll get detailed information about reported injuries, malfunctions, and deaths tied to specific equipment, ranging from pacemakers to IV pumps. This tool lets you apply granular filters: you can restrict the search by a device name, filter by a precise date range, or narrow results down based on the type of event that occurred.

Because these three tools—query_drug_events, query_food_recalls, and query_medical_devices—all accept raw Lucene syntax inputs, you can run complex cross-domain queries. If your research requires comparing drug events with a specific recall date range, or checking device malfunctions against documented pathogen outbreaks, the server handles that heavy lifting. Your AI client acts like it's sitting at the terminal running the query; it sends the exact search logic and gets back the structured data payload from the live source.

This setup means you bypass typical developer overhead. You don't need to manage API key rotation or build out complex, multi-step data pipelines just to get raw public health information. Your agent simply executes the specialized query against the massive government datasets and returns what you asked for. It’s built for researchers who demand unfiltered access: whether you need to monitor outbreaks in a specific state, check side effects reported across multiple drugs, or track malfunctions on a device type over a ten-year span, this server gets the data flowing directly to your analysis.