Healthie (BYOK)
Read your Healthie appointment types, appointments (without patient details), availability and programs with your own API key
Drug facts from the FDA label, not from memory
Unify Dynamics LLC
Real questions, and the tools your agent calls to answer them.
drug_lookuplabel_sectionside_effectsinteractionssimilar_drugsrecallsshortagesndc_lookupclinical_trialshealth_topicMeds answers drug and health questions from the U.S. primary sources and never from memory. A name is normalised with RxNorm (brand ↔ generic, misspellings fixed), then the drug's current FDA-approved label is read: what it is approved for, how it is taken, boxed warning, contraindications, side effects, interactions, pregnancy, pediatric and geriatric use, overdose, storage, mechanism, patient information — any section, in full, with the DailyMed link. Side effects come two ways: what the label lists and what has actually been reported to FDA's FAERS system, with counts and the share marked serious. The interaction check reads both labels and quotes every sentence where one names the other — by drug, brand or class — and says 'not listed' rather than 'safe' when neither does. Also: the other drugs in the same class, FDA recalls with lots and reasons, the shortage list with availability and resupply estimates, what an NDC on a bottle is, recruiting clinical trials by condition and place, and MedlinePlus health topics in plain language. Keyless, read-only, nothing stored. Information, not medical advice.
Sources: openFDA (FDA labels, NDC directory, enforcement reports, drug shortages, FAERS), RxNorm and MedlinePlus (U.S. National Library of Medicine), ClinicalTrials.gov (NIH)
The publisher, Unify Dynamics LLC, states that tool inputs are processed in memory and not retained.
10 tools, grouped by job. Every one is read-only or compute-only.
https://meds.connectorbay.com/mcpRead your Healthie appointment types, appointments (without patient details), availability and programs with your own API key
Read your Cliniko practice setup, practitioners, appointment types, appointment times, billable items and products with your own API key
Read the daily physical, sleep, body and nutrition summaries of the wearable users in your ROOK project with your own client credentials
Search FDA drug adverse event reports, drug labels, and drug, food and device recalls with BYOK
Auto-enrolment earnings trigger, qualifying earnings band and minimum 3% employer and 5% employee contributions for 2026-27, for any pay period
900+ exercises with instructions, workout plans, calories burned, BMI/TDEE, macros, 1RM, HR zones, pace
https://meds.connectorbay.com/mcpPOST https://meds.connectorbay.com/v1/tools/{tool_name}, described by the OpenAPI 3.1 spec at https://meds.connectorbay.com/openapi.jsonAuthorization: Bearer <key> raises rate limits.https://meds.connectorbay.com/health · Tool schemas: https://meds.connectorbay.com/tools.jsonMost "what is this drug" answers on the internet are a paraphrase of a paraphrase. Meds goes to the documents themselves: the FDA-approved label (via openFDA) for what a drug is for, how it is taken, its warnings, contraindications, side effects and interactions; RxNorm for what a name actually is (brand ↔ generic, ingredients, forms); FAERS for what has been reported to FDA once the drug was on the market; the FDA recall and shortage lists; the NDC directory for the code on the bottle; ClinicalTrials.gov for studies; and MedlinePlus for plain-language health topics. Every answer links to the source it came from.
A name is normalised with RxNorm first (so "metformn" → metformin, "Lipitor" → atorvastatin), then the best current label is chosen: the product named that, from its original manufacturer, most recent version first; a single-ingredient oral product over a combination or an injection when the name is a generic; a repackager's copy only as a last resort (the approved text is the same). Section headings and table noise are stripped, and long sections are clipped with the full length reported so the agent can ask for more. OTC "Drug Facts" labels have no overdose or pregnancy section — Meds pulls the sentences from Warnings that address the topic and says that is what it did.
There is no free, keyless interaction database any more (NLM retired its interaction API), so interactions does the
one thing that can be done from primary sources: it reads both labels and reports every sentence in which one names the
other — by generic, brand, or class ("NSAIDs", "CYP3A4 inhibitors", "ACE inhibitors"). A hit is quoted verbatim; no hit
is reported as not listed, never as safe, with a nudge to the pharmacist.
Meds is information, not medical advice, and it never invents a dose. FAERS counts are voluntary reports, not rates. Dietary supplements, non-U.S. brands and compounded products are outside the FDA label set and are said to be. For a possible overdose or poisoning the answer is always Poison Control (1-800-222-1222) or 911.
drug_lookup — What a drug is, from RxNorm and its current FDA label: brand ↔ generic, active ingredients, drug class, Rx or OTC, routes and dose forms, manufacturer, approval number, what it is approved for (label 'uses'), the boxed warning if any, which label sections exist, links to the DailyMed label and the MedlinePlus consumer page. Start here for any drug question; use label_section for the full text of a section. Args: name, prefer.
label_section — The full text of one section of the drug's FDA-approved label — uses, dosage, forms_and_strengths, boxed_warning, warnings, contraindications, side_effects, interactions, pregnancy, pediatric, geriatric, overdose, storage, mechanism, pharmacokinetics, clinical_studies, patient_info, ingredients, abuse_and_dependence, and OTC 'ask_doctor' / 'stop_use'. Use for "how is X taken", "who must not take X", "what does the label say about pregnancy". Args: name, section, prefer, max_chars, set_id.
side_effects — Two views of a drug's adverse effects: what the FDA label lists (adverse reactions section, and the boxed warning if any) and what has actually been reported to FDA's FAERS system in the last N years — the most-reported reactions with counts and the share of reports marked serious. FAERS is voluntary reporting: counts are signals, not rates. Use for "side effects of X", "is hair loss reported with Y". Args: name, prefer, years.
interactions — Do two drugs' FDA labels warn about each other? Searches each label's interaction, warning and contraindication sections for the other drug — by generic and brand name and by drug class (e.g. 'NSAIDs', 'CYP3A4 inhibitors') — and returns the matching label sentences. 'Not listed' means neither label mentions the other, which is NOT the same as safe: tell the user to confirm with a pharmacist. Use for "can I take X with Y". Args: drug_a, drug_b.
similar_drugs — Other drugs in the same FDA pharmacologic class (EPC) as this one — e.g. every statin, every SSRI — as generic names with how many labels each has, plus the class's mechanism-of-action term. Use for "what are the alternatives to X", "which drugs are in the same class as Y". Alternatives are not interchangeable; a prescriber chooses. Args: name, prefer, limit.
recalls — FDA drug recalls (enforcement reports): product, reason, lots, firm, class (I = could cause serious harm), distribution and dates, newest first. Use for "any recalls for X", "eye-drop recalls this year", "Class I drug recalls this month". Args: query, days, classification, limit.
shortages — FDA's drug shortage list: which presentations are in shortage, availability per manufacturer, reason, estimated resupply and the company contact. Use for "is X in shortage", "why can't I get Y", "when will Z be back". Args: query, status, limit.
ndc_lookup — Identify a product from its NDC: brand and generic name, strength, form, route, labeler, Rx/OTC, DEA schedule, package description, marketing category and the label link. Use for "what is NDC 0071-0155-23", "what's in this bottle". Args: ndc.
clinical_trials — Studies on ClinicalTrials.gov by condition, intervention, status and place: title, phase, sponsor, status, enrollment, eligibility (age, sex), sites, central contact and the summary, with links. Use for "trials for X recruiting near Y", "is there a phase 3 for Z", "who is studying W". Args: condition, intervention, status, location, keywords, limit.
health_topic — Plain-language health information from MedlinePlus (U.S. National Library of Medicine): the topic summary — what it is, causes, symptoms, diagnosis, treatment — with the page link, related names and MeSH terms. Use for "what is atrial fibrillation", "what causes kidney stones", "what does an A1C test measure". Args: query, limit.
ok: true with data, or ok: false with a structured error (bad_input, not_found, upstream_error). degraded lists optional sources that were unreachable.Retry-After. initialize and tools/list are never counted.agent@connectorbay.com · connectorbay.com/meds/support
Effective 22 September 2026. Unify Dynamics LLC ("we", "us"), Menlo Park, California.
What Meds is. Meds is a connector (an MCP server and REST API) that AI assistants and personal agents call on your behalf for U.S. drug and health information: FDA labels, side effects, label-based interaction checks, recalls, shortages, NDC lookup, clinical trials and MedlinePlus health topics. You interact with your assistant; the assistant sends us only the specific request needed: a drug name, an NDC, a condition or a topic.
What we receive. The arguments of each tool call, the technical headers needed to serve the request (an IP address or a client identifier supplied by the assistant, and a Pro API key if you have one), and standard server logs.
Health information is sensitive; we keep none of it. A request naming a drug or a condition could reveal something about a person's health. We do not store request arguments: they are used to fetch the answer and discarded. We do not receive your conversation with the assistant, your name, contacts, health records or credentials for any service. Rate-limit counters (a client identifier and timestamps) are held in memory for at most 24 hours. Hosting-provider access logs, which contain IP addresses and request paths but not request bodies, are retained for up to 30 days for security and abuse prevention. Nothing is used to build a profile, sold, shared, or used to train models.
Nothing personal is needed. Meds answers questions about drugs and conditions in general; it never asks who is taking a medicine, for a date of birth, a prescription number or any identifier.
Sub-processors. To answer requests we send only the drug name, code, condition or topic asked about (never anything about you) to openFDA (U.S. Food and Drug Administration), RxNorm and MedlinePlus (U.S. National Library of Medicine) and ClinicalTrials.gov (U.S. National Institutes of Health). Hosting: Vercel (United States regions).
Not a covered entity. Meds is a general-information service and does not provide care, hold medical records or act as a HIPAA covered entity or business associate.
Legal bases and your rights. For users in California, the EU/UK and elsewhere, our processing is necessary to provide the service you request through your assistant. Because we store no personal data beyond short-lived logs, requests to access, correct or delete personal data are usually satisfied automatically; you can still contact agent@connectorbay.com and we will respond within 30 days.
Children. Meds is a general-audience service operated by the adult using the assistant.
Changes. We will post changes here with a new effective date.
Contact. Unify Dynamics LLC · agent@connectorbay.com
Effective 22 September 2026. These terms govern use of the Meds connector ("the Service") operated by Unify Dynamics LLC ("we", "us").
interactions reports what two labels say about each other. A result of "not listed" means neither label mentions the other, not that the combination is safe.Email agent@connectorbay.com. We answer within two business days.
https://meds.connectorbay.com/mcphttps://meds.connectorbay.com/openapi.jsonWhen you report a problem, include the question you asked your agent, roughly when, and which agent you used (for example Claude, ChatGPT or Gemini). Never send us passwords, account numbers or ID numbers. Meds never asks for them.