wunder-user-mcp
An MCP server that exposes the Wunder Mobility v2 end-user API as tools, enabling clients to browse vehicles, manage rentals, and create reservations on behalf of a signed-in user.
README
wunder-user-mcp
An MCP server that exposes the Wunder Mobility v2 end-user (User) API as tools. It lets an MCP client (Claude, etc.) browse vehicles, inspect and operate rentals, and create rentals/reservations on behalf of a single signed-in end user.
Authentication is handled entirely from a refresh token you provide via env var — the server exchanges it for short-lived access tokens and refreshes them automatically, so tool callers never deal with tokens.
Tools
| Tool | What it does | Underlying endpoint |
|---|---|---|
get_vehicles |
List available vehicles. With a latitude/longitude, annotates each vehicle with distance from the user and sorts ascending. |
GET /vehicles |
get_active_rental |
Return the user's ongoing rental(s) (ACTIVE / RESERVATION). |
GET /rentals |
rental_command |
Operate a rental: START, PARK, DRIVE, END, OPEN_TAILBOX, RENEW_RESERVATION, … |
POST /rentals/{id}/operation |
create_rental |
Create a rental in RESERVATION or ACTIVE state from a vehicle_id or vehicle_code. |
POST /rentals |
Configuration
All configuration is via environment variables (see .env.example):
| Variable | Required | Default | Purpose |
|---|---|---|---|
WUNDER_REFRESH_TOKEN |
✅ | — | Long-lived refresh token for the end user. |
WUNDER_TENANT |
✅ | — | Tenant short-code used in the API URL path. |
WUNDER_BASE_URL |
https://go.api.gourban.services/v1 |
API gateway base URL. | |
WUNDER_BRANCH_ID |
— | Default branch id for vehicle/rental calls. | |
WUNDER_DEFAULT_RADIUS_MI |
5 |
Radius (miles) used by get_vehicles when a location is given without a radius. |
|
WUNDER_HTTP_TIMEOUT |
30 |
Per-request timeout in seconds. |
Getting a refresh token
Sign in once via the Wunder Authentication API (e.g. POST /{tenant}/auth/sign-in-email,
/sign-in-phone-number, or /sign-in-api-client) and copy the refreshToken from the
response into WUNDER_REFRESH_TOKEN.
Run
Run straight from the GitHub repo — no clone and no PyPI publish required (uvx builds
from the Git source):
uvx --from git+https://github.com/slettmayer/wunder-user-mcp wunder-user-mcp
Pin a branch, tag, or commit for stability by appending @<ref>:
uvx --from git+https://github.com/slettmayer/wunder-user-mcp@main wunder-user-mcp
uvx caches the build, so to pick up new commits on a moving ref (e.g. @main) add
--refresh:
uvx --refresh --from git+https://github.com/slettmayer/wunder-user-mcp@main wunder-user-mcp
From a local checkout (development):
uv run wunder-user-mcp # editable install, reflects local edits
uvx --from . wunder-user-mcp # build from the current directory
In all cases WUNDER_REFRESH_TOKEN and WUNDER_TENANT must be set in the environment
(see Configuration). The server speaks MCP over stdio.
MCP client configuration
{
"mcpServers": {
"wunder-user": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/slettmayer/wunder-user-mcp",
"wunder-user-mcp"
],
"env": {
"WUNDER_REFRESH_TOKEN": "<refresh-token>",
"WUNDER_TENANT": "<tenant-code>",
"WUNDER_BASE_URL": "https://go-staging.api.gourban.services/v1",
"WUNDER_BRANCH_ID": "<optional-branch-id>"
}
}
}
}
WUNDER_BASE_URL defaults to production (https://go.api.gourban.services/v1); the
value above targets staging (go-staging). Set it to match your tenant's environment
— a staging tenant on the production host returns 404. To pin a ref, append @<ref> to
the URL, e.g. git+https://github.com/slettmayer/wunder-user-mcp@main.
Development
uv sync # install deps (incl. dev group)
uv run pytest # offline unit tests (geo + JWT exp parsing)
uv run ruff check # lint
Notes / caveats
- The vehicle position field path used for distance calculation is detected
best-effort (
lat/lng,latitude/longitude, or a nestedposition/location). If a tenant's response uses different field names, extendgeo.py. operation_typeinrental_commandis passed through to the API, so tenant-specific operations beyond the documented set keep working.- Rental creation and
ENDare high-impact: they can trigger vehicle commands, payment authorization/capture, deposits, and invoices.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。