generated: '2026-09-02'
method: searched
source: https://medblocks.com/docs/mcp
name: medblocks-platform
title: Medblocks Platform MCP Server
description: 'Hosted, remote MCP server that gives an AI assistant access to a Medblocks Platform workspace:
  orient in the workspace, search health systems, start a patient-portal connection session, read the
  FHIR records that come back, and disconnect or delete.'
deployment:
  mode: remote
  endpoint: https://app.medblocks.com/mcp
  auth: oauth
  verified: probed
  note: 'POST tools/list to https://app.medblocks.com/mcp with Accept application/json, text/event-stream
    returned HTTP 401 with WWW-Authenticate: Bearer resource_metadata="https://app.medblocks.com/.well-known/oauth-protected-resource/mcp",
    scope="openid profile email offline_access patients:read patients:write patient_sessions:read patient_sessions:write
    connections:read fhir:read". That is a conformant RFC 9728 challenge, so the endpoint is live and
    the live tool schemas require an authenticated introspection. Developer clients may substitute an
    mb_sk_ API key as a bearer token instead of the OAuth flow.'
  probe_prior: (never probed)
  probe: gated
  probe_why: RFC 9728 challenge on the MCP path only
  checked: '2026-09-11'
  source: claimed-backlog re-probe 2026-09-11
x-evidence:
  fetched: '2026-09-02'
  url: https://app.medblocks.com/mcp
  method: POST
  body: '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
  http_status: 401
  www_authenticate: Bearer resource_metadata="https://app.medblocks.com/.well-known/oauth-protected-resource/mcp",
    scope="openid profile email offline_access patients:read patients:write patient_sessions:read patient_sessions:write
    connections:read fhir:read"
transport: streamable-http
clients_documented:
- name: Claude Code
  install: claude mcp add --transport http medblocks https://app.medblocks.com/mcp
- name: Codex
  install: codex mcp add medblocks --url https://app.medblocks.com/mcp --oauth-resource https://app.medblocks.com/mcp
  note: the --oauth-resource flag is required; the server declines a token not bound to this resource
- name: claude.ai / Cowork
  install: Settings > Connectors > Add custom connector, URL https://app.medblocks.com/mcp
- name: ChatGPT
  install: Developer mode > Plugins > New plugin, Server URL https://app.medblocks.com/mcp, Authentication
    OAuth
consent:
  model: An interactive OAuth user is shown a consent screen with one tick box per capability, all ticked
    by default, and may grant fewer permissions than the assistant requested. Identity (name, email) and
    workspace visibility are granted without a prompt.
  permissions:
  - label: Connect your healthcare facilities
    allows: search named hospitals, clinics and insurers, connect to them, and bring records into the
      workspace
    default: true
  - label: Read your health records
    allows: read the health records already in the workspace
    default: true
  - label: Manage people and delete data
    allows: add or remove people, disconnect facilities, delete records
    default: true
    note: destructive; the docs advise leaving it off unless needed
  revocation: Dashboard > Settings > MCP server revokes the assistant; already-issued access expires within
    an hour.
tools:
  count: 10
  status: published
  schemas: auth-gated
  note: Names and descriptions are taken verbatim from the provider's own MCP documentation and from the
    provider-published agent skill (skills/medblocks-mcp.md). Input schemas were not captured because
    tools/list requires an authenticated token. The docs state every tool call writes a private audit
    entry, which is why even lookup tools declare that they change state in their MCP metadata.
  list:
  - name: get_workspace_overview
    description: Reports who you are, summarizes the selected workspace, and lists health systems previously
      connected there.
    category: orientation
  - name: list_workspaces
    description: Lists every workspace you can use, labeled production or sandbox.
    category: orientation
  - name: list_people
    description: Lists everyone set up in your workspace.
    category: orientation
  - name: search_health_systems
    description: Searches hospitals, clinics, and insurers by name.
    category: discovery
  - name: choose_health_systems
    description: Opens the in-chat panel with every matching facility so you can make the selection.
    category: discovery
  - name: connect_health_system
    description: Creates a private connection session and secure link after you select one or more facilities.
      Reuses an existing person with the same name or creates a new person from the real name you provide.
    category: connect
  - name: check_connection_status
    description: Checks how a connection is going, for a single session or for a whole person.
    category: connect
  - name: get_health_records
    description: Reads a person's medical records from their connected health systems.
    category: read
  - name: disconnect_health_system
    description: Stops retrieving records from one connected health system and revokes access.
    category: destructive
    destructive: true
  - name: delete_person
    description: Permanently removes a person and all their retrieved records from the workspace.
    category: destructive
    destructive: true
skills_over_mcp:
  published: true
  note: The server publishes the Medblocks agent skill through the draft Skills-over-MCP extension; ChatGPT
    reads the catalog on Scan Tools. Clients without the extension see ordinary tools and resources.