Genefoundry

io.github.berntpopp由 GitHub 账号 berntpopp 发布v0.13.17更新于 Nov 22, 2026

MCP gateway federating 30 biomedical MCP servers behind one endpoint: gnomAD, ClinVar, HPO, VEP.

已验证Streamable HTTP可网页运行知识与记忆AI 与机器学习

概览

AI 生成的概览

远程 MCP 网关,将 gnomAD、ClinVar、HPO、VEP 等 22 个生物医学数据服务器聚合到单一端点,并支持按意图搜索工具。

可以做什么
  • 按意图在 22 个生物医学后端、约 285 个工具的联合目录中搜索
  • 用带命名空间的名称调用命中的工具,例如 gnomad_search_genes 或 spliceai_predict_splicing
  • 直接访问各后端固定的入口工具,不依赖搜索排序
  • 通过一个连接查询变异、基因、表型、蛋白、表达、药物基因组学与文献数据源
适用场景
适合需要查询遗传学与生物医学参考数据的场景,例如变异频率与临床意义、基因-疾病注释、表型本体、蛋白功能、剪接预测,而无需逐个接入各数据源。
运行要求
需要 genefoundry.org 的远程 Streamable-HTTP 端点,无需本地安装;必须使用 OAuth 认证。自行部署需 Python 3.14+ 与 uv,并配置 GF_* 后端地址与认证模式。
安装前请注意
明确仅供科研使用,不构成临床决策支持,不得用于诊断、治疗、分诊或患者管理。调用方令牌不会转发至上游,但查询仍会发往第三方生物医学数据源。工具定义会与已审核基线做指纹比对以检测漂移。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Genefoundry,将其添加到工作区。
  2. 连接 所需的 凭据。它们会被私密存储,绝不会在公开页面上展示。
  3. 为需要使用其工具的对话启用该服务。

Web executable,通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体,或 SourceWeft 对话中。智能体会读取该服务器的安装指南,先向你展示发布者、端点和登录方式,经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/io.github.berntpopp%2Fgenefoundry/install.md,并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前,先告诉我它的发布者、端点和登录方式,等我确认。不要把 API 密钥写进命令或我会提交的文件;问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。 客户端会在首次连接时完成 OAuth 登录。

{
  "mcpServers": {
    "genefoundry": {
      "type": "http",
      "url": "https://genefoundry.org/mcp"
    }
  }
}

README

genefoundry-router

[Python 3.14+] [CI] [Security] [License: MIT]

A thin FastMCP 3.x aggregator that federates the GeneFoundry *-link MCP fleet behind a single Streamable-HTTP endpoint. A host adds one server — genefoundry — and gets every biomedical backend with collision-free <namespace>_<tool> naming and search-based discovery.

[!IMPORTANT] Research use only. Not clinical decision support. Do not use for diagnosis, treatment, triage, or patient management.

Why

An MCP host that mounted all 21 backends directly would face a wall of several hundred tools — more than a model can reason over, and a guarantee of name collisions. The router collapses that into one endpoint and replaces the flat catalog with a search surface, so a model finds the right tool by intent instead of by scrolling.

It is a client to each backend and a server to hosts: it namespaces and shapes the surface, but never rewrites a backend's data. The caller's token is never forwarded upstream.

Quick start

The fleet is hosted — no install required:

bash
claude mcp add --transport http genefoundry https://genefoundry.org/mcp

Health check: genefoundry.org/health.

To run your own against the live fleet (Python 3.14+, uv):

bash
uv sync --group devcp .env.example .env                    # set GF_*_URL backend URLs and GF_AUTH_MODEuv run genefoundry-router run --host 127.0.0.1 --port 8000curl -s localhost:8000/health | python -m json.tool

An offline fake fleet (make dev-fleet + make run-dev, or one-shot make test-e2e) runs the real router against impersonated backends over real Streamable-HTTP — no Docker, no network.

Tools

The router does not surface the federated catalog flat. A model sees three things:

ToolPurpose
search_toolsRelevance search over the entire federated catalog
call_toolInvoke a hit by its <namespace>_<tool> name
pinned entry pointsEach backend's front-door tool, always visible — declared per-backend as entrypoints: in servers.yaml
text
search_tools(query="splicing prediction")   # → spliceai_predict_splicing (+ schema)call_tool(name="spliceai_predict_splicing", arguments={...})

Pinning makes each domain's canonical tool reachable deterministically rather than by relevance luck. See How discovery works — including the two traps that bite MCP clients.

Federated backends

30 backends, 345 tools, each surfaced namespaced — e.g. gnomad_search_genes.

NamespaceDomainData sourceToolsRepo
pubtatorLiterature & entity annotationPubTator335pubtator-link
gnomadVariant / gene / population frequencygnomAD22gnomad-link
hpoPhenotype ontology & associationsHuman Phenotype Ontology19hpo-link
orphanetRare disease ontology & associationsOrphadata19orphanet-link
clingenGene–disease curationClinGen17clingen-link
mavedbVariant-effect assay scoresMaveDB15mavedb-link
uniprotProtein functionUniProt15uniprot-link
clinpgxPharmacogenomics / gene-drug guidelinesClinPGx13clinpgx-link
genereviewsGene–disease literatureGeneReviews13genereviews-link
mgiMouse phenotype & modelsMGI13mgi-link
mondoDisease ontology / cross-referencesMondo13mondo-link
genccGene–disease curationGenCC12gencc-link
metadomeProtein tolerance landscapesMetaDome11metadome-link
civicSomatic variant clinical evidence & AMP/ASCO/CAP assertionsCIViC10civic-link
stringdbProtein–protein interaction networksSTRING10stringdb-link
cbioportalCancer genomics mutation prevalence & hotspotscBioPortal9cbioportal-link
gtexTissue expressionGTEx Portal9gtex-link
hgncGene nomenclatureHGNC9hgnc-link
panelappDiagnostic gene panels & curationPanelApp9panelapp-link
clinvarVariant clinical significanceClinVar8clinvar-link
pgscatalogPolygenic scores, traits, ancestry & performance metricsPGS Catalog8pgscatalog-link
ucscGenome Browser region tracks, reference sequence, BLAT & in-silico PCRUCSC Genome Browser8ucsc-link
autopvs1Variant ACMG PVS1AutoPVS17autopvs1-link
opentargetsTarget tractability, drug candidates & target-disease evidenceOpen Targets Platform7opentargets-link
spliceaiSplicing predictionSpliceAI Lookup7spliceailookup-link
vepVariant annotation / consequenceEnsembl VEP7vep-link
litvarVariant literatureLitVar26litvar-link
moalmanacRegulatory approvals for cancer biomarker-therapy pairsMolecular Oncology Almanac5moalmanac-link
strchiveTandem-repeat (STR) disease loci & repeat rangesSTRchive5strchive-link
mitomapMitochondrial DNA (mtDNA) variants & disease associationsMITOMAP4mitomap-link

Data & provenance

The router serves no data of its own; each backend owns its sources, licences and citation guidance, and the router mirrors their disclaimers.

What it does own is integrity of the tool surface. A backend can serve a clean tool at review time and later change its definition — the channel for a rug pull. The router fingerprints every normalized tool definition and diffs the live fleet against a reviewed, packaged baseline (genefoundry_router/data/fleet-baseline.json), enforced at startup and on a schedule. See Deployment → drift detection.

Documentation

Contributing

See AGENTS.md for engineering conventions. make ci-local is the definition-of-done gate: format, lint, line budget, README standard, mypy, and tests.

License

MIT © Bernt Popp. Each federated backend carries the licence and citation terms of its upstream data source; see that backend's repository.

来源:README.md,提交 e59d0e8

工具

0
该服务器需要登录,因此 SourceWeft 无法在没有账号的情况下读取其工具,也未收录。已连接账号所能看到的内容可能因账号及其权限而异。

版本历史

23
  1. v0.13.17最新Nov 22, 2026
  2. v0.13.15Nov 21, 2026
  3. v0.13.14Nov 21, 2026
  4. v0.13.13Nov 20, 2026
  5. v0.13.12Nov 20, 2026
  6. v0.13.11Nov 19, 2026
  7. v0.13.7Nov 19, 2026
  8. v0.13.2Nov 19, 2026
  9. v0.12.3Nov 19, 2026
  10. v0.10.17Nov 13, 2026
  11. v0.10.13Nov 12, 2026
  12. v0.10.12Nov 11, 2026
  13. v0.10.10Oct 29, 2026
  14. v0.10.8Oct 21, 2026
  15. v0.10.7Oct 20, 2026
  16. v0.10.4Oct 18, 2026
  17. v0.10.1Oct 4, 2026
  18. v0.10.0Oct 4, 2026
  19. v0.9.4Oct 4, 2026
  20. v0.9.3Oct 3, 2026
  21. v0.9.2Sep 18, 2026
  22. v0.9.1Sep 18, 2026
  23. v0.9.0Sep 16, 2026

评分

已满足适用检查项的 85%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通必需25 分
  • 已声明认证方式必需10 分
  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分