User MCP Server
Streamable HTTP MCP server that exposes user and billing management tools, resources, and prompts through a REST API, enabling natural language interaction with user profiles, bills, and sharing features.
README
User MCP Servers
Streamable HTTP MCP Server:通过 HTTP 调用独立的 User REST API(my-user-service),覆盖 MCP 三要素(Tools / Resources / Prompts)。
本仓库 不包含 FastAPI / SQLite / JWT 实现。MCP 只依赖 API_BASE_URL。
架构
┌──────────────────────┐ HTTP /mcp ┌──────────────────────┐
│ MCP Client │ ──────────────────────────▶ │ MCP Server :3001 │
│ (Cursor / Inspect) │ │ Streamable HTTP │
└──────────────────────┘ └──────────┬───────────┘
│ httpx
▼
┌──────────────────────┐
│ User REST API :8000 │
│ (my-user-service) │
└──────────────────────┘
快速开始
先启动 API(另一个仓库):
cd /Users/jasqia/00D_PythonProject/my-user-service
source .venv/bin/activate
uvicorn app.main:app --host 127.0.0.1 --port 8000 --reload
再启动 MCP:
cd /Users/jasqia/00D_PythonProject/my-mcp-servers
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # 可选;默认 API_BASE_URL=http://127.0.0.1:8000
python -m mcp_server
或一条命令拉起两边(API 目录默认同级的 ../my-user-service):
./scripts/start_services.sh
# 自定义 API 仓库路径:
# USER_SERVICE_ROOT=/path/to/my-user-service ./scripts/start_services.sh
- MCP 端点:http://127.0.0.1:3001/mcp
- REST 文档:http://127.0.0.1:8000/docs
端口被占用(Address already in use)
启动时如果看到:
ERROR: [Errno 48] error while attempting to bind on address ('127.0.0.1', 3001): [errno 48] address already in use
说明 3001 已被占用(拆仓后常见原因:旧仓库 my-mcp 里残留的 python -m mcp_server 还在跑)。
- 查出占用进程:
lsof -nP -iTCP:3001 -sTCP:LISTEN
输出里的 PID 就是占用端口的进程号,例如:
COMMAND PID USER FD TYPE ...
Python 83386 jasqia ... TCP 127.0.0.1:3001 (LISTEN)
- 确认是哪个项目(可选):
ps -p <PID> -o pid,command=
- 停掉该进程后重试启动:
kill <PID>
# 若未退出:kill -9 <PID>
python -m mcp_server
- 确认端口已空闲:
lsof -nP -iTCP:3001 -sTCP:LISTEN || echo "port 3001 is free"
如果不想停旧进程,也可以换端口:
MCP_PORT=3002 python -m mcp_server
Cursor / Inspect 需改连 http://127.0.0.1:3002/mcp。API :8000 被占用时,处理方式见 my-user-service README。
MCP 三要素
Tools
用户
register_user/login_user/get_current_user/update_user_profile
账单
create_bill/list_my_bills/list_shared_bills/get_billupdate_bill/delete_billshare_bill/unshare_billlike_bill/unlike_bill
Resources
user://api/health— API 健康检查user://docs/overview— 项目说明user://profile/{access_token}— 当前用户资料bill://mine/{access_token}— 我的账单列表bill://shared/{access_token}— 分享给我的账单bill://item/{access_token}/{bill_id}— 单条账单
Prompts
welcome_new_user— 欢迎新用户help_update_profile— 协助更新资料help_create_bill— 协助记账help_share_bill— 协助分享账单help_like_bill— 协助点赞
Cursor 连接(Streamable HTTP)
{
"mcpServers": {
"user-mcp": {
"url": "http://127.0.0.1:3001/mcp"
}
}
}
本地验证
source .venv/bin/activate
pip install -r requirements-dev.txt # 首次需要
# 单元 / 协议测试(不需要先起 API)
make test
# make test-e2e # L4 真 Streamable HTTP
# 手工 E2E 冒烟(需 API + MCP 都已启动)
python scripts/test_mcp_user_client.py
python scripts/test_mcp_bill_client.py
测试分层说明:docs/MCP_L2_L4_EXPLAINED.zh.md
Inspect 评测
先启动 API + MCP,再:
pip install -r requirements-inspect.txt
export GOOGLE_API_KEY='...'
./scripts/run_inspect_gemini.sh google/gemini-3.6-flash evals/bill_mcp_scenarios.py
./scripts/run_inspect_gemini.sh google/gemini-3.6-flash suite
# 无 API Key:用 mockllm 本地验证 Inspect→MCP 接线
python scripts/run_inspect_user_mcp_smoke_mock.py
目录结构
my-mcp-servers/
├── mcp_server/ # Streamable HTTP MCP Server
│ ├── config.py # MCP_HOST / MCP_PORT / API_BASE_URL
│ ├── __main__.py # python -m mcp_server
│ └── server.py # Tools / Resources / Prompts
├── tests/mcp/
├── evals/ # Inspect 评测
├── scripts/
├── docs/
├── Makefile
├── pytest.ini
├── requirements.txt
└── README.md
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.