medicover-mcp
Enables AI agents to manage appointments on Medicover's Polish patient portal, including booking, rescheduling, and slot searching for patients and dependents.
README
medicover-mcp
An MCP server that lets an AI agent manage appointments on
Medicover's Polish patient portal (https://online24.medicover.pl/). It talks directly to the
portal's private JSON API (reverse-engineered from real traffic — see
NOTES.md); it does not drive a browser at runtime.
It is built for an unattended daily routine: hunt for scarce slots and optionally rebook an existing appointment to an earlier one — including for linked dependents (e.g. children).
Tools
| Tool | Purpose |
|---|---|
list_patients |
The account holder + linked dependents, each with a stable patient_id (MRN). |
list_appointments |
A patient's planned/past appointments (read-only). |
search_slots |
Available slots by region + specialty (optionally doctor/clinic/date). |
book_appointment |
Book a slot for a patient — patient_id required. |
reschedule_appointment |
Move an appointment to another slot; "earlier-only" by default. |
list_appointment_filters |
Resolve specialty/doctor/clinic/region names → IDs from history. |
list_search_filters |
Clinics/doctors/languages bookable for a region + specialty. |
Patient-scoped read tools default to the account holder; the write tools require an
explicit patient_id so a booking can never land on the wrong person. Nothing else (messaging,
records, prescriptions, payments, cancel-without-rebook) is exposed.
Authentication (two-phase)
Login uses SMS 2FA, which can't be in an unattended hot path, so:
-
One-time bootstrap (interactive CLI, not a tool):
export MEDICOVER_USERNAME=<your-MRN-or-login> export MEDICOVER_PASSWORD=<your-password> # or omit to be prompted uv run medicover-auth login # prompts for the SMS codeThis completes OAuth + 2FA and saves the refresh token to a
0600file (default~/.config/medicover-mcp/credentials.json; override withMEDICOVER_STORE_PATH). Username/password are never stored. -
Unattended runtime: the MCP server loads the refresh token and silently refreshes the short-lived (~3 min) access token on every run. No password or SMS is ever needed during tool calls.
-
On expiry/revocation: tools fail with
session expired — run medicover-auth login; they never prompt for an SMS through a tool call.
Token-lifetime decision: the in-MCP
request_sms_code/submit_sms_codefallback is not shipped — see the reasoning inNOTES.md.
Helpers: uv run medicover-auth status (shows whether creds exist, no secrets) and
uv run medicover-auth logout (deletes them).
Setup
uv sync # create the venv and install
uv run medicover-auth login
uv run pytest # run the tests (no network — HTTP is mocked)
MCP client config
Add to your MCP client (e.g. Claude Desktop claude_desktop_config.json):
{
"mcpServers": {
"medicover": {
"command": "uv",
"args": ["--directory", "/abs/path/to/medicover", "run", "medicover-mcp"],
"env": {
"MEDICOVER_STORE_PATH": "/home/you/.config/medicover-mcp/credentials.json"
}
}
}
}
(Run medicover-auth login once on the same machine first so the credential store exists.)
Environment variables
| Var | Used by | Meaning |
|---|---|---|
MEDICOVER_USERNAME / MEDICOVER_PASSWORD |
bootstrap only | Login credentials (else prompted). |
MEDICOVER_STORE_PATH |
bootstrap + server | Credential file path (default ~/.config/medicover-mcp/credentials.json). |
MEDICOVER_TIMEOUT |
server | HTTP timeout seconds (default 30). |
Example agent flow (earlier-slot hunt for a child)
list_patients→ find the child'spatient_id.list_appointments(patient_id=child, state="Planned")→ note the appointment to improve and itsappointment_id/date.list_appointment_filters(patient_id=child)→ get the region id and specialty id.search_slots(patient_id=child, region_ids=[...], specialty_ids=[...], start_date="today").- If an earlier slot exists:
reschedule_appointment(patient_id=child, appointment_id=..., booking_string=<slot>, new_slot_date=<slot date>)(refuses unless strictly earlier).
Safety
Read-before-write on booking/reschedule (the portal validates the slot at action time); reschedule never cancels without rebooking and defaults to earlier-only; polite request pacing with backoff; tokens/credentials are never logged or returned; request/response bodies (which carry medical data) are not logged.
Disclaimer
Unofficial, built against an undocumented private API that may change without notice. Use with your own account and within Medicover's terms. Not affiliated with Medicover.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.