Skip to main content
PracticeHub is a practice management platform for healthcare clinics. The PracticeHub API is the single, canonical way to work with an account’s data from outside the app: a REST API for integrations and automations, and an MCP server so AI assistants can act inside a clinic on a team member’s behalf.

REST API

Patients, appointments, practitioners, locations, invoices, payments, files and more — read, filter, create, update, delete.

MCP Server

Connect Claude or ChatGPT to a clinic. Same data, same rules, discovered automatically — read-only during the limited rollout.

Docs MCP

This site is an MCP server too. Point Claude Code, Cursor or any AI tool at it and install the skill file — it reads documentation only, never a clinic.

How it is put together

  • One account, one tenant. Every clinic has its own subdomain and its own isolated database. Your base URL is https://{clinic}.{region}.practicehub.io/v3/api — the host the practice signs in on (the region host is the one in the practice’s login URL — neptune (London), mercury (Frankfurt), vulcan (Ohio), jupiter (Singapore) or apollo (Sydney)) — and a credential only ever reaches the account it was created on.
  • Resources, not endpoints. Each resource (patients, appointments, …) is a uniform collection: list with filters, fetch one, create, update, delete. GET /schema tells you exactly what every resource exposes and accepts.
  • The app’s rules apply. Writes follow the same rules the PracticeHub app enforces — a booking checks the patient is active and the resource is at that location, a payment updates the balance, voiding an invoice releases its allocations. You get field-level validation and meaningful status codes instead of surprises.
  • Your data on our records. Every record carries a metadata object for your own identifiers, which you can filter on.
  • Push, not poll. Webhooks — live in a limited rollout — tell you when records change — whether the change came through the API, from staff in the app, or from a background job — with thin, signed payloads.
  • Built to be discovered. The OpenAPI reference on this site is generated from the running API, and the MCP server exposes the same model to AI clients.

Where to next

1

Authenticate

Create an API key and make your first call — Authentication.
2

Query

Filters, sorting, pagination and discovery — Querying.
3

Write

Creating and changing records safely — Writing.
4

Build

The best practices proven integrations follow. Stuck? FAQ.