convez.ai
MCP ServerIntegração & APIsAvançado
Cadasto

openEHR Assistant MCP Server

Cadasto/openehr-assistant-mcp

MCP server para modelagem clínica openEHR com integração a IA

Ver no GitHub
BomPluginScore · de 0 a 100

Servidor MCP que facilita o trabalho com openEHR — archetypes, templates, AQL, terminologia e especificações — oferecendo integração direta com Claude, Cursor e outros clientes. Inclui ferramentas para navegação do Clinical Knowledge Manager, resolução terminológica e validação de sintaxe ADL.

Instalação

$npx convez-ai Cadasto/openehr-assistant-mcp

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
18
Forks
8
Acompanhando
2
Saúde
62%
Último commit
há 3 semanas
Versão
0.21.0
Issues abertas
5
Contribuidores
2

Casos de uso

  • Exploração e validação de archetypes openEHR
  • Assistência em design e revisão de templates clínicos
  • Navegação e documentação do CKM (Clinical Knowledge Manager)
  • Consultas e análise semântica de terminologia médica

Veredito

Ferramenta especializada para profissionais que trabalham com padrões openEHR em saúde digital. Bem estruturada como servidor MCP com suporte a múltiplos clientes e deploy flexível (hospedado ou local), mas requer conhecimento prévio de openEHR e ADL para ser útil.

README

A tradução para o português deste README ainda está na fila.

The openEHR Assistant MCP Server

PR validation Release Docker image (GHCR) License: MIT PHP Version MCP

An MCP server that helps AI assistants work with openEHR — archetypes, templates, AQL, terminology, and specifications.

Working with openEHR means navigating the Clinical Knowledge Manager (CKM), intricate type systems, and ADL syntax rules. This server gives MCP clients (Claude Desktop, Cursor, LibreChat, …) direct access to those resources so assistants can help with archetype exploration, semantic explanation, language translation, syntax correction, and design reviews.

Pre-release: expect frequent updates and breaking changes until version 1.0.

Recommended: pair this server with the openEHR Assistant Plugin — skills, prompts, and agents that guide assistants through openEHR workflows. Claude Code users can install it from the Cadasto Plugin Marketplace.

Table of Contents


Features

  • Works with any MCP client (Claude Desktop, Cursor, LibreChat, …).
  • Tools, prompts, resources, and completions for openEHR archetypes, templates, AQL, terminology, and specifications.
  • Guided prompts orchestrate multi-step modelling and review workflows.
  • Use the hosted endpoint, or run locally (transports: streamable HTTP and stdio).

Quick Start

The quickest path is the hosted endpoint — point your MCP client at it:

URLhttps://openehr-assistant-mcp.apps.cadasto.com/
Transportstreamable-http
{
  "mcpServers": {
    "openehr-assistant-mcp": {
      "type": "streamable-http",
      "url": "https://openehr-assistant-mcp.apps.cadasto.com/"
    }
  }
}

To run your own instance (Docker or stdio) and for per-client setup (Claude Desktop, LibreChat, Cursor, Junie), see docs/install.md.


Available MCP Elements

Tools

CKM (Clinical Knowledge Manager)

  • ckm_archetype_search — List Archetypes from CKM matching search criteria
  • ckm_archetype_get — Get a CKM Archetype by its identifier
  • ckm_template_search — List Templates (OET/OPT) from CKM matching search criteria
  • ckm_template_get — Get a CKM Template (OET/OPT) by its identifier

openEHR Terminology

  • terminology_resolve — Resolve a terminology concept ID to its rubric, or find the ID for a given rubric across groups

Guides (model-reachable)

  • guide_search — Search bundled guides and return short snippets with canonical openehr://guides URIs
  • guide_get — Retrieve guide content by URI or (category, name), chunked by default
  • guide_adl_idiom_lookup — Look up targeted ADL idiom snippets for common modelling patterns

Examples (curated artefacts)

  • examples_search — Search bundled examples (AQL, FLAT/STRUCTURED payloads, ADL archetypes) and return snippets with openehr://examples URIs
  • examples_get — Retrieve an example by URI or (kind, name)

openEHR Type specification

  • type_specification_search — List bundled openEHR Type specifications matching search criteria
  • type_specification_get — Retrieve an openEHR Type specification (as BMM JSON)

Prompts

Optional prompts that guide AI assistants through common openEHR and CKM workflows using the tools above.

  • ckm_archetype_explorer — Discover and fetch CKM Archetype definitions (ADL/XML/Mindmap)
  • ckm_template_explorer — Discover and fetch CKM Template definitions (OET/OPT)
  • type_specification_explorer — Discover and fetch openEHR Type specifications (BMM JSON)
  • terminology_explorer — Discover and retrieve openEHR terminology (groups and codesets)
  • guide_explorer — Discover and retrieve openEHR implementation guides
  • explain_archetype — Explain an archetype's semantics (audiences, elements, constraints)
  • explain_template — Explain openEHR Template semantics
  • explain_aql — Explain an AQL query's intent, structure, and semantics
  • explain_simplified_format — Explain context, paths, and data elements of a FLAT/STRUCTURED payload
  • translate_archetype_language — Translate an archetype's terminology section between languages with safety checks
  • fix_adl_syntax — Correct or improve ADL syntax without changing semantics; before/after + notes
  • design_or_review_archetype — Design or review an archetype for a concept/RM class, with structured output
  • design_or_review_template — Design or review an openEHR Template (OET)
  • design_or_review_aql — Design or review an AQL query, using the AQL guides
  • design_or_review_simplified_format — Design or review a FLAT/STRUCTURED instance, using the Simplified Formats guides

Completion Providers

Parameter suggestions in MCP clients when invoking tools or resources.

  • Guides — guide {name} values per category (openehr://guides/{category}/{name})
  • Examples — example {name} values per kind (openehr://examples/{kind}/{name})
  • SpecificationComponents — {component} values from resources/bmm (openehr://spec/type/{component}/{name})

Resources

Exposed via #[McpResource] and fetchable by clients using openehr://… URIs.

  • Guides — openehr://guides/{category}/{name} (Markdown). Categories: archetypes, templates, aql, simplified_formats, specs (per-document spec digests), howto (toolchain how-tos). Retrieve via guide_search / guide_get.
    • e.g. openehr://guides/aql/principles, openehr://guides/specs/rm-ehr, openehr://guides/howto/spec-lookup
  • Examples — openehr://examples/{kind}/{name}. Kinds: aql, flat, structured (Markdown: metadata header + fenced code block), archetypes (native .adl, text/plain). Retrieve via examples_search / examples_get.
    • e.g. openehr://examples/aql/latest_blood_pressure_per_ehr, openehr://examples/archetypes/openEHR-EHR-OBSERVATION.blood_pressure.v2
  • Type Specifications — openehr://spec/type/{component}/{name} (BMM JSON).
    • e.g. openehr://spec/type/RM/COMPOSITION, openehr://spec/type/AM/ARCHETYPE
  • Terminology — openehr://terminology (JSON): all openEHR terminology groups and codesets.

Documentation

  • docs/install.md — hosted & local setup, client configurations
  • docs/development.md — Docker dev environment, Makefile, configuration, MCP Inspector
  • docs/conventions.md — coding standard and MCP authoring conventions
  • docs/testing.md — tests, static analysis, MCP conformance
  • docs/ — the Specification-Driven Development spec set (requirements, architecture, decisions, traceability)
  • CONTRIBUTING.md — how to contribute · AGENTS.md — repository instructions for AI coding agents

Contributing

Contributions are welcome — see CONTRIBUTING.md for setup, conventions, and how to propose changes, and CHANGELOG.md for notable changes.

License: MIT — see LICENSE.


Acknowledgments

This project is inspired by and grateful to:

Ver o original no GitHub

Bom para

Relacionados