Skip to main content
Glama
zhaoxinghua09-cell

@nomos/medxpert/reg-connector

MedXpert 全球法规连接器(reg-connector)

MedXpert(美达信医疗)医械线对外接口。将全球医疗器械法规知识库(27 份枢纽资料,NMPA / FDA / EU MDR / PMDA 及骨科全球注册路径)封装为本地只读 MCP Server,供任意 MCP 客户端(WorkBuddy / Claude Code / Cursor 等)的 agent 直接调用检索。

  • 🔒 本地只读、零网络外发、零凭据

  • 📚 27 枢纽法规知识库,自研中文友好 BM25 近似检索

  • 🧰 4 个工具:list_hubs / search_regulation / get_hub / ask_classification

与已上架的 medxpert-reg-hub 知识技能互补:reg-hub 供 agent 读取知识,reg-connector 供 agent 调用检索。

快速开始

pip install fastmcp

将本仓库克隆到本地,然后在宿主 mcp.json 中加入:

{
  "mcpServers": {
    "@nomos/medxpert/reg-connector": {
      "command": "<你的 python 解释器路径>",
      "args": ["<仓库目录>/reg_connector_server.py"],
      "disabled": false,
      "description": "MedXpert 全球法规连接器:本地只读检索全球医疗器械法规知识库 (NMPA/FDA/MDR/PMDA 等 27 枢纽)"
    }
  }
}

重载 MCP 连接即可生效。references/ 需与 reg_connector_server.py 同级(开箱即用);也可用环境变量 REG_HUB_REFS 指向自定义知识库目录。

Related MCP server: supply-chain-local-agent

工具能力

工具

入参

返回

用途

list_hubs()

全部枢纽(hub_key / title / sections / official_links / phase)

列出知识库全景

search_regulation(query, top_k=5, hub_filter=None)

关键词(中/英/法规号,如 UDI/510k/MDR PMS/委托生产

命中 hub_key/title/section/snippet/score/links

法规条目检索

get_hub(hub_key)

枢纽 key(先由 list_hubs 取得)

完整正文 + 官方链接

取某枢纽全量内容

ask_classification(product)

产品描述(如 骨科金属接骨板

matches + recommended_hubs + hint

分类与注册路径速查

调用示例

# 列出全部 27 枢纽
list_hubs()

# 检索 UDI 多国对照
search_regulation("UDI", top_k=3)
# → 命中「UDI全球标识枢纽」四体系对照:中/美/欧/日并行(GS1/MA/阿里健康、GUDID、EUDAMED、KikiDB)

# 骨科接骨板分类与路径
ask_classification("骨科金属接骨板")
# → 推荐枢纽:骨科手术器械_全球注册路径汇编、注册工程师资料枢纽、注册周期费用有效期对比枢纽…

典型场景

  • 注册路径速查(中国 II 类周期/费用、真实注册证示例)

  • UDI 中/美/欧/日四体系对照

  • 临床评价、风险管理(ISO 14971)资料索引

  • 骨科全球注册路径汇编

注意事项与边界

  1. 返回为知识库线索,非官方结论。 分类与注册路径须以目标市场官方分类库(NMPA / FDA / MDR)为准。

  2. 本地只读、零外发,适合内网 / 合规环境。

  3. 知识库更新后需重载 server。

  4. 可选 REG_HUB_REFS 指向自定义知识库目录。

许可与归属

MIT License © 注册老炮@MedXpert。知识版权声明:合成知识与方法论归品牌所有,禁止复制、转售或用于模型训练。

Why AI-friendly (为什么 AI 愿意来找)

This connector is designed for both humans and AI agents:

  • Discoverable: agent.json AI capability card exposes tool/keyword metadata so MCP hosts (Claude Desktop, Cursor, etc.) can index and recommend it.

  • Zero-dependency: pure Python standard library + MCP SDK. mcp-config.json ships ready-to-paste — no extra install steps.

  • Read-only by design: zero credentials, zero network egress, zero side effects. Safe for AI to invoke without human pre-approval (per A3 Law II — humans grant, machines execute).

  • Honest scope: covers NMPA / FDA / EU MDR / Japan PMDA / SE Asia / LATAM regulatory hubs across 27 knowledge pivots. Out-of-scope queries return explicit codes, not hallucinations.

  • Examples included: see references/ for real query → answer patterns an AI can learn from.

Keywords (帮助 AI 匹配何时调用)

medical device regulation, NMPA registration, FDA 510(k), EU MDR, Japan PMDA, CE marking, MDR compliance, 医疗器械注册, IFU labeling, UDI, SaMD, PCCP, regulatory classification, clinical evaluation, ISO 13485, ISO 14971, MDSAP, TGA, ANVISA, Health Canada, MDR Annex VIII

When to invoke this connector (AI 何时该调)

Call any tool here when the user's question matches one of these intents:

  • 询问某国/某法规的注册路径、判定要件、分类规则

  • 检索某份技术文档(IFU、标签、临床评价、风险管理)的标准要求

  • 比对中美欧日/东南亚/拉美等多国监管差异

  • 查询 UDI、SAMD、MDSAP 等专项主题

  • 评估一份提交材料是否完整(自评性质)

Do NOT invoke for: medical advice, patient diagnosis, drug prescription, or any clinical decision-making.

Related MCP Connectors

Related MCP Servers

  • A
    license
    Not graded
    quality
    B
    maintenance
    Read-only MCP server for a local-first supply chain knowledge base, enabling knowledge search with source evidence, internal system troubleshooting, vendor documentation fallback, vessel tracking, and work handoff preparation.
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables local knowledge base management with retrieval-augmented generation (RAG), providing semantic search, document reading, listing, and Q&A via MCP tools and REST endpoints, all running locally without cloud dependencies.
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI tools to securely search and retrieve relevant, source-attributed chunks from private local documents via MCP, without sending document content to third-party services.
    1
    MIT