onehome-mcp

onehome-mcp

Enables searching curated OneHome property listings, fetching property details and photos, comparing homes, and running mortgage calculations from within Claude.

Category
访问服务器

README

onehome-mcp

CI npm license

MCP server for OneHome (CoreLogic) — search the listings your real-estate agent curated for you, fetch property details + photos, compare houses side-by-side, and run mortgage / affordability math from within Claude.

Sister project to zillow-mcp, redfin-mcp, compass-mcp, and homes-mcp. Same tool ergonomics — different upstream auth model.

This project was developed and is maintained by AI (Claude). Use at your own discretion.

What's different about OneHome

OneHome isn't a public listings site. Buyers usually reach it through a magic link an agent emails them — https://portal.onehome.com/...?token=eyJ.... That token query param IS the per-user bearer that the portal SPA hands to every GraphQL request.

So instead of routing every fetch through your signed-in browser tab (like the other realty MCPs), onehome-mcp talks directly to services.onehome.com/graphql from Node, with Authorization: Bearer <jwt> attached. We support three ways to source that bearer:

Mode How to enable Notes
env_token ONEHOME_TOKEN=<jwt> Paste the raw bearer from devtools Network panel. Most direct.
magic_link ONEHOME_MAGIC_LINK=https://portal.onehome.com/...?token=... Paste the full URL your agent sent — we extract the token param.
fetchproxy_capture (no env) + fetchproxy extension installed + signed-in portal.onehome.com tab We wait for your tab to fire any GraphQL request, snapshot the Authorization header, and reuse it.

Tools

Tool What it does
onehome_get_user Smallest auth probe — returns your OneHome profile (name, email) and the groups your agent shared.
onehome_get_groups List the OneHome "groups" your agent has shared with you (each one a market / curated listing bucket).
onehome_get_saved_search Fetch an agent-curated saved search by id — name, filter criteria, polygon, and the OSK listing ids that compose the share.
onehome_get_saved_search_with_listings The "show me my saved homes" flow in one round trip — saved search plus its inflated listings.
onehome_search_properties Listings within a group; optionally scoped to a saved search.
onehome_search_suggestions Free-text suggestion search (address, MLS #) across all feeds.
onehome_get_by_address Resolve a single free-text street address to a listing's portal URL + id.
onehome_resolve_addresses Bulk-resolve up to 100 structured addresses to portal URLs + listing ids; concurrent, per-row error capture.
onehome_get_property Full property record by listing id or portal URL.
onehome_bulk_get Fetch up to N listings in one call — one structured row per id, per-row error capture.
onehome_get_property_photos Full media gallery — Thumbnail / Medium / Large variants + room descriptions.
onehome_compare_properties 2-8 listings side-by-side. Per-row error capture; calls are concurrent.
onehome_get_schools Local-Logic primary + high schools near a lat/lng.
onehome_get_walk_score Local-Logic walk / transit / bike / car friendliness scores.
onehome_graphql Power-user escape hatch — send a raw GraphQL document with variables.
onehome_calculate_mortgage Local PITI calculator. Same math as the other realty MCPs.
onehome_calculate_affordability Local 28/36 DTI solver — max home price you can afford.
onehome_set_auth Add another authenticated session at runtime (magic link / JWT / email-token) for buyers holding shares across multiple agents.
onehome_set_active_session Force a specific registered session to be the active one (overrides MLS-suffix routing).
onehome_get_session_context List every registered session — auth mode, token expiry, and the group / saved-search / agent scope each bootstrapped.
onehome_healthcheck End-to-end auth + GraphQL smoke check with token-expiry diagnostics.

Install

The simplest path is the published Claude plugin (.mcpb install). For local dev:

git clone https://github.com/chrischall/onehome-mcp
cd onehome-mcp
npm install
npm run build

Then point your MCP host at node /abs/path/to/onehome-mcp/dist/bundle.js with one of:

// claude_desktop_config.json
{
  "mcpServers": {
    "onehome-mcp": {
      "command": "node",
      "args": ["/abs/path/to/onehome-mcp/dist/bundle.js"],
      "env": {
        "ONEHOME_MAGIC_LINK": "https://portal.onehome.com/en-US/properties/map?token=eyJ..."
      }
    }
  }
}

Development

npm test               # vitest, mocked transport, no network
npm run test:watch
npm run test:coverage
npx tsc --noEmit
npm run build          # tsc --noEmit + esbuild → dist/bundle.js

Tests use a FakeTransport (in tests/helpers.ts) that registers per-operationName handlers — there's no live network in the test suite. The tests/index.test.ts smoke check loads the same tool registrations src/index.ts uses against an in-memory MCP client/server pair, so "I wrote the tool file but forgot to wire it up" mistakes fail loudly.

License

MIT.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选