About CUQU找搭子 MCP（粗趣）
CUQU找搭子 MCP（粗趣） gives AI agents access to CUQU, a local offline-activities platform in China, for participants and organizers. It covers discovery through booking, payment, and check-in, separating public queries from keyed write actions.
Key features
- 01Search activities by type, city, and date
- 02View activity details and venue listings
- 03Generate activity sharing links
- 04Publish activities and register attendees
- 05Create WeChat Pay orders and check in
Use cases
- 01Find local interest-based activities
- 02Review event times, venues, and booking progress
- 03Publish an offline event as an organizer
- 04Handle booking, payment, and on-site check-in
MCP support
It offers a hosted Streamable HTTP endpoint and can also run over stdio. MCP clients can use public read tools; write actions such as publishing, registration, payment, and check-in run through the hosted gateway and require a user API key.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://agent.cuqu.net/mcpclaude mcp add --transport http cuqu-net-cuqu https://agent.cuqu.net/mcp{"mcpServers":{"cuqu-net-cuqu":{"type":"http","url":"https://agent.cuqu.net/mcp"}}}Open in Cursor{"servers":{"cuqu-net-cuqu":{"type":"http","url":"https://agent.cuqu.net/mcp"}}}Open in VS Code{"mcpServers":{"cuqu-net-cuqu":{"type":"http","url":"https://agent.cuqu.net/mcp"}}}{"mcpServers":{"cuqu-net-cuqu":{"serverUrl":"https://agent.cuqu.net/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
8| Parameter | Type | Required | Description |
|---|---|---|---|
| date | string | No | |
| limit | integer | No | Maximum number of results to return, 1–100, default 50. Pass a larger value to see more. |
| keyword | string | No | |
| district | string | No | District/county name, e.g. Nanshan, Futian, Longgang, Huidong County. When passed together with city, filters by city and district in two levels; common aliases (area/zone) are supported. Omit to skip district filtering. |
| location | string | No | |
| activity_type | string | No |
Details
From the project READMECUQU找搭子 (CUQU) MCP Server
Open-source MCP server giving AI agents live access to CUQU (找搭子) — a local offline social activities platform in China (Shenzhen-based, 10+ cities incl. Hong Kong, China). Users discover and book offline interest-based group activities (board games, frisbee, hiking, badminton, fishing, anime and 100+ more categories); organizers publish and run events.
Live demo endpoint (hosted): https://agent.cuqu.net/mcp (Streamable HTTP, per-user cq-sk- keys — see llms.txt)
Agent Card: https://cuqu.net/.well-known/agent.json
Talk to the agents (chat UI):
- Baidu Qianfan: https://appbuilder.baidu.com/s/w93d7Z6X (Chinese)
- iFlytek Astron: https://agent.xfyun.cn/agentbuilder/chat?sharekey=998b5236b21fb85e005b9c9438f78b10&botId=5830291 (Chinese; API flow_id
7510896245077958656) - Live weekly activities: https://cuqu.net/activities/list/ (real in-stock data, updated daily)
- IRL Activity Schema v0.1 (CC BY 4.0): https://cuqu.net/schema/irl-activity/v0.1/irl-activity-schema.json
What this repository contains
| Path | What it is |
|---|---|
server.py | The MCP server implementation — zero-dependency Python, JSON-RPC 2.0 over stdio. Read tools call CUQU's public live activity API; write tools forward to the hosted gateway with a user key. |
Dockerfile | One-command container build for the server (glama.ai / docker / any MCP host). |
glama.json | glama.ai directory metadata & maintainership. |
scripts/check-mcp.py | Connectivity check for the hosted gateway. |
skills/, docs/ | Agent skill manifests and platform integration docs (Qwen/Dify/Yuanqi). |
Tools
8 tools across the activity lifecycle:/n
| Tool | Type | Description |
|---|---|---|
query_activities | read (no key) | Search live activities by type / city / date — 100+ categories, real-time public data |
get_activity_card | read (no key) | Full detail card for one activity (time, venue, price, booking progress, organizer, cover) |
query_venues | read (no key) | Curated partner venue directory (capacity, per-head price) |
generate_urllink | read (no key) | Share link + WeChat mini-program path for an activity |
create_activity | write (key) | Organizer publishes a new event |
register_event | write (key) | Book a user into an activity |
create_payment | write (key) | Create WeChat Pay for a booking order |
check_in | write (key) | On-site check-in / redemption |
Write tools are executed by the hosted gateway and require CUQU_API_KEY (cq-sk-..., issued via scan-to-issue flow documented in llms.txt).
Quick start
# stdio, zero dependencies, Python 3.9+
python -u server.py
# or via Docker
docker build -t cuqu-mcp .
docker run -i cuqu-mcp
Claude Desktop / any stdio MCP client:
{
"mcpServers": {
"cuqu": {
"command": "python",
"args": ["-u", "/path/to/server.py"],
"env": { "CUQU_API_KEY": "cq-sk-...(optional, enables write tools)" }
}
}
}
License
MIT — © 2026 光天科技（深圳）有限公司 (Guangtian Technology (Shenzhen) Co., Ltd.)
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.