Skip to main content
Glama
deenesjakoruzh
Krv-Labs

real-world-mcp

by Krv-Labs
OverviewSchemaRelated ServersScoreDiscussions
Python
Hybrid

真实世界证据与临床可行性 MCP 服务器

仓库： Krv-Labs/how-to-train-your-dragon

基于纵向索赔数据（Komodo）和公共注册库（ClinicalTrials.gov），对 Graves 病方案（NCT07570316）按每 12 个月预期随机化受试者数对美国医疗机构进行排名，并以交互式 Model Context Protocol (MCP) 服务器形式提供给 AI 智能体使用。

Python 包名为 httyd；MCP 服务器注册名为 real-world-mcp。CLI 入口点：httyd、realworld-mcp 和 how-to-train-your-dragon。

为什么叫这个名字？ 基于 Komodo Health 索赔数据构建——将 Komodo 龙驯服并训练成可操作的可行性模型。

1. 架构与包结构

how-to-train-your-dragon/
├── httyd/                          # Main Python package
│   ├── models/                     # L1–L5 parameters, provenance registry, recruitment math
│   ├── analysis/                   # Scoring pipeline, catchment, accrual, site linkage
│   ├── data/                       # Dataset catalog, loaders, CTG fetch
│   ├── evaluations/                # Temporal holdout validation
│   ├── auth/                       # Kubernetes token auth & scope enforcement
│   ├── engine.py                   # In-memory FeasibilityEngine singleton
│   ├── cache.py                    # Precomputed site/geo cache loader
│   └── server.py                   # MCP server (16 tools, 4 resources, 2 prompts)
├── ui/                             # Local observation dashboard (offline HTML)
│   ├── build_dashboard.py
│   └── (writes to output/site_feasibility.html)
├── data/                           # Claims & registry assets (not committed — see below)
├── docs/                           # Methodology, scoring review, evidence
├── deploy/helm/httyd/      # GKE Helm chart
├── scripts/precompute.py           # Build startup cache (Docker + local dev)
├── tests/
├── pyproject.toml
└── README.md

数据说明： Komodo 索赔 CSV 已被 gitignore。仅提交 data/nct07570316_sites.json（真实基准数据）。运行引擎前，请将授权数据文件放在本地 data/ 目录中。

方法论与验证证据：docs/README.md。

2. 快速开始

安装

uv sync

预计算缓存（推荐）

全量全国评分重建约需 90 秒。预计算一次后即可复用：

uv run python scripts/precompute.py
export REALWORLD_CACHE_DIR=./cache   # optional; defaults to ./cache when present

运行测试

REALWORLD_CACHE_DIR=./cache uv run pytest

部分测试（敏感性重新评分、验证）会触发全量重建，总计约需 2 分钟。

启动 MCP 服务器

# stdio transport (default — Claude Desktop, Cursor)
uv run httyd

# streamable-http (local dev)
AUTH_ENABLED=false REALWORLD_CACHE_DIR=./cache uv run httyd streamable-http
curl http://localhost:8080/health

生成本地预览仪表盘

uv run python ui/build_dashboard.py
open output/site_feasibility.html

3. MCP 工具面

real-world-mcp 在 6 个领域共暴露 16 个工具：

A. 数据集发现与检查

  • list_datasets、describe_dataset、get_cohort_summary

B. 模型注册与溯源

  • list_models、get_provenance

C. 中心搜索与深度分析

  • search_and_score_sites、get_site_details、explain_site_score

D. 空间覆盖范围与篮子优化

  • compute_basket_catchment、find_nearby_competitors

E. 入组预测与敏感性分析

  • simulate_accrual_timeline、update_interim_accrual、compute_sensitivity_rescore、match_trial_facility

F. 验证与局限性

  • run_model_validation、get_model_limitations

4. MCP 智能体配置

{
  "mcpServers": {
    "real-world-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/how-to-train-your-dragon",
        "run",
        "httyd"
      ],
      "env": {
        "REALWORLD_CACHE_DIR": "/path/to/how-to-train-your-dragon/cache"
      }
    }
  }
}

5. GKE 内部部署

作为内部 ClusterIP MCP 服务部署。子智能体 Pod 使用 Kubernetes ServiceAccount 令牌进行认证，并通过 TokenReview API 验证。

架构

  • 传输方式： streamable-http，端口 8080，路径 /mcp

  • 认证： Bearer 令牌 = 投影的 ServiceAccount 令牌（audience: httyd）

  • 授权： ConfigMap 将 system:serviceaccount:<ns>:<name> 映射到作用域

  • 网络： NetworkPolicy 仅允许来自标记为 httyd-client: "true" 的命名空间的入站流量

作用域

访问权限

feasibility:read

发现、搜索、解释、溯源工具

feasibility:simulate

入组模拟、敏感性重新评分、篮子覆盖范围

构建与部署

docker build -t httyd:0.1.0 .

helm upgrade --install httyd deploy/helm/httyd \
  -n realworld --create-namespace \
  --set image.repository=REGION-docker.pkg.dev/PROJECT/REPO/httyd \
  --set image.tag=0.1.0

部署前，请在 deploy/helm/httyd/scopes.yaml 中注册子智能体 ServiceAccount。

子智能体 Pod 配置

kubectl label namespace app httyd-client=true
serviceAccountName: subagent-feasibility-reader
volumes:
  - name: mcp-token
    projected:
      sources:
        - serviceAccountToken:
            audience: httyd
            expirationSeconds: 3600
            path: token
volumeMounts:
  - name: mcp-token
    mountPath: /var/run/secrets/tokens
    readOnly: true
env:
  - name: HTTYD_MCP_URL
    value: "http://httyd.httyd.svc:8080/mcp"
  - name: HTTYD_MCP_TOKEN_FILE
    value: "/var/run/secrets/tokens/token"

环境变量

变量

默认值

用途

AUTH_ENABLED

false（本地）/ true（容器）

启用 K8s 令牌认证

MCP_RESOURCE_URL

http://httyd.httyd.svc:8080

OAuth 资源标识符

MCP_SCOPE_CONFIG

/etc/mcp/scopes.yaml

SA → 作用域映射

MCP_TOKEN_AUDIENCE

httyd

预期令牌受众

REALWORLD_CACHE_DIR

./cache

预计算的中心/地理 JSON

PORT

8080

HTTP 监听端口

HOST

0.0.0.0

HTTP 绑定地址

STATELESS_HTTP

true

无状态 MCP 会话（K8s 推荐）

This server cannot be installed

-
license - not tested
Not graded
quality - not tested
C
maintenance

How are these scores calculated?

Maintenance

–Maintainers
–Response time
–Release cycle
–Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

View all MCP Connectors

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Krv-Labs/how-to-train-your-dragon'

If you have feedback or need assistance with the MCP directory API, please join our Discord server