Skip to main content
Glama

deviCERTS medical device regulatory data

Search FDA recalls

search_recalls
Read-onlyIdempotent

Search U.S. FDA medical device recalls across all devices by words (product description, recall reason, recalling firm or recall number), product code, firm, submission number or pathway, start-date range, recall class and status, with offset paging. Returns up to 20 recalls per page, newest first: recall number, class, status, start date, reason, product code, firm, submissions and the name-matched DevicErts manufacturer page, plus total and next_offset. Use when the question covers recalls in general or all recalls of a company, e.g. ongoing Class I recalls of a product code. Do not use for the recalls of one known device (use get_device_recalls) or for EU data (recalls are FDA only). Read-only, no login. Data: public EUDAMED and U.S. FDA records, synced monthly, EU certificates weekly (dates in get_data_sources). Cached up to 1 hour; limit 300 calls a minute per IP.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "search_recalls"
  }
}

Input Schema

NameRequiredDescriptionDefault
firmNoOnly recalls whose recalling firm name contains these words, 2 to 100 characters. e.g. "Medtronic". Default: all.
limitNoResults per page, integer 1 to 20. Default 10.
queryNoWords in the product description, recall reason, recalling firm or recall number, 2 to 200 characters. e.g. "infusion pump", "battery", "Z-0079-2022". Default: none.
offsetNoNumber of results to skip, integer 0 to 999. Default 0. Pass next_offset from the previous page to get the next one; offset plus limit stays within the first 1000 results.
statusNoOnly recalls with this FDA status: ongoing (open), completed or terminated. Default: all.
date_toNoOnly recalls started on or before this date, YYYY-MM-DD. e.g. 2024-12-31. Default: no upper bound.
pathwayNoOnly recalls naming a submission of this FDA premarket pathway: 510k, pma or denovo. Default: all.
date_fromNoOnly recalls started on or after this date, YYYY-MM-DD. e.g. 2024-01-01. Default: no lower bound.
product_codeNoOnly recalls of this three-letter FDA product code. e.g. DXN, FRN. Default: all.
recall_classNoOnly recalls of this FDA recall class: Class I (most serious), Class II or Class III. About a third of recalls have no class yet and never match. Default: all.
submission_numberNoOnly recalls naming this FDA submission: a 510(k), PMA, De Novo or HDE number. e.g. K132592, P180028. Default: all.

Output Schema

NameRequiredDescriptionDefault
totalYes
formatYes
offsetYes
cite_asYes
resultsYes
next_offsetYes
retrieved_atYes
status_as_ofYes
canonical_urlYes

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A4.4/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond the annotations (readOnly, idempotent, non-destructive), it discloses ordering (newest first), the 20-per-page cap, total/next_offset in the response, monthly/weekly sync cadence and a 1-hour cache, and a rate limit of 300 calls/minute per IP — genuinely useful operational context. Minor deductions: the returned field list largely duplicates the output schema, and the line 'Data: public EUDAMED and U.S. FDA records' sits awkwardly beside 'recalls are FDA only', which could briefly confuse the agent about source scope.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with purpose, then returns, then when-to-use/when-not, then operational data — a sensible order with no filler sentences. It is dense and a little long, and the enumeration of returned fields repeats what the output schema already states.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For an 11-parameter, zero-required search tool with a full output schema, the description supplies everything an agent still needs: which corpus, result ordering, page size, paging handoff, cache and rate limits, and the boundary with the sibling device-recall tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents every parameter including examples, enums and the note that many recalls lack a class. The description adds only modest extra meaning — that results are newest-first and that next_offset drives the following page — which is close to the baseline for a fully documented schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The first sentence gives a specific verb and resource (search, U.S. FDA medical device recalls) plus the exact dimensions searchable (words, product code, firm, submission number/pathway, dates, class, status) and the paging model. It explicitly contrasts scope with get_device_recalls, so an agent can separate the two without inspecting either schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It states a positive trigger (questions about recalls in general or all recalls of a company, with a concrete example) and two explicit exclusions with alternatives: one known device goes to get_device_recalls, and EU data is out of scope. No inference is required to choose this tool.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.