China 12306 Train Tickets MCP

China 12306 Train Tickets MCP

[License: MIT] [MCP] [Remote]

Query China Railway 12306 ticket availability from any MCP client — train schedules and remaining seats by Chinese station name and date. Read-only, no booking, no account.

• Try it in 30 seconds: a free public MCP endpoint is already running — just paste the URL into your MCP client (no install, no API key).
• Or self-host: a single Python file, stdlib HTTP + FastMCP, zero paid dependencies.

⚡ Use the hosted endpoint (no setup)

https://mcp.pianam.cn/train-mcp/mcp

Transport: Streamable HTTP (MCP 2025-03-26 compatible). No authentication required.

🔌 Client configuration

Add this to your MCP client's mcpServers configuration (Claude Desktop claude_desktop_config.json, Cursor mcp.json, Cline, Cherry Studio, etc.):

{  "mcpServers": {    "12306-train": {      "type": "http",      "url": "https://mcp.pianam.cn/train-mcp/mcp"    }  }}

Clients that do not accept "type": "http" (some Cherry Studio / older Cline versions) accept the same entry with just "url".

🧰 Tools

📡 Data sources, caching & limits

• Data comes from 12306's own public query endpoints (kyfw.12306.cn/otn/leftTicket/...) — read-only queries only; this server cannot book tickets, please buy via the official 12306 app/website.
• Station code table cached locally for 7 days; ticket queries cached for 300 seconds.
• No API key, no login required.
• The hosted endpoint is rate-limited to 200 requests / minute / IP.

🐢 Self-hosting

git clone https://github.com/boy-373/12306-train-mcp.gitcd 12306-train-mcppip install -r requirements.txtpython train_mcp_server.py# the server listens on 127.0.0.1:8003 by default; override with:#   MCP_HOST=0.0.0.0 MCP_PORT=9000 python train_mcp_server.py#   MCP_ALLOWED_HOSTS="your-domain.com,127.0.0.1:*"#   MCP_ALLOWED_ORIGINS="https://your-domain.com"

Then point your MCP client at http://127.0.0.1:8003/mcp. No API keys or accounts are ever required.

🐳 Self-host with Docker

Option A — pull the pre-built image (fastest, no build):

docker run -d -p 8000:8000 --name 12306-train-mcp ghcr.io/boy-373/12306-train-mcp:latest

Option B — build from source:

git clone https://github.com/boy-373/12306-train-mcp.gitcd 12306-train-mcpdocker build -t 12306-train-mcp .docker run -d -p 8000:8000 --name 12306-train-mcp 12306-train-mcp

Then point your MCP client at http://127.0.0.1:8000/mcp. The container listens on 0.0.0.0:8000 by default (override with -e MCP_PORT=9000 and adjust -p accordingly). No API keys or accounts required.

• train_stations.json — bundled 12306 station name → telegraph-code table (auto-refreshed from 12306 every 7 days).

🗂️ Files

• train_mcp_server.py — the MCP server (FastMCP, Streamable HTTP transport).
• rate_limit.py — lightweight per-IP sliding-window rate-limit middleware (200 req/min default).
• requirements.txt — mcp, uvicorn, starlette.
• server.json — official MCP Registry manifest (remote server entry, ready to publish with mcp-publisher).
• smithery.yaml / glama.json — directory listing metadata.

🧯 Parameter guide & common errors (read this if a call fails)

query_train_tickets(from_city, to_city, date="")

Why you may get an HTTP 400 / error result:

1. Bad date format — must be exactly YYYY-MM-DD (2026/09/20, 2026年9月20日, Sep 20 all fail).
2. Station name not found — typo, pinyin ("qingdao"), or a vague nickname the code table doesn't contain. Fix: call search_station with a keyword and copy the exact name.
3. City-level name for a multi-station city — e.g. "北京" usually resolves to 北京 station, but if you want a specific station pass "北京南" / "北京西" explicitly.
4. Date outside the 15-day pre-sale window — 12306 has not released those tickets yet.

Correct call examples:

query_train_tickets(from_city="青岛", to_city="北京", date="2026-09-20")query_train_tickets(from_city="北京南", to_city="上海虹桥")        # date omitted -> tomorrowsearch_station(keyword="上海")                                   # -> 上海/上海南/上海虹桥/上海西…

A successful response returns 车次总数 plus a 车次 list (train number, departure/arrival station & time, duration, and remaining seats per class). An unrecognized station returns {"error": "出发站「…」无法识别，可用 search_station 工具查车站名"} — simply re-call with the exact name.

🧯 参数说明与常见错误（调用失败先看这里）

query_train_tickets(from_city, to_city, date="") 参数说明

返回 400 / 错误结果的常见原因：

1. 日期格式错误：必须严格为 YYYY-MM-DD，2026/09/20、2026年9月20日、Sep 20 等写法都会失败；
2. 站名不存在：错别字、拼音（如 "qingdao"）、或码表中没有的俗称都会无法识别。解决方法：先用 search_station 关键词搜索，复制返回的准确站名；
3. 多车站城市用了城市通称：如 "北京" 一般能匹配到「北京」站，但要去特定车站请明确传 "北京南"、"北京西" 等；
4. 日期超出 15 天预售期：12306 尚未放票，查询会返回失败提示。

正确调用示例：

query_train_tickets(from_city="青岛", to_city="北京", date="2026-09-20")query_train_tickets(from_city="北京南", to_city="上海虹桥")        # date 留空 -> 默认明天search_station(keyword="上海")                                   # 返回 上海/上海南/上海虹桥/上海西…

调用成功会返回 车次总数 和 车次 列表（车次号、出发/到达站与时间、历时、各席别余票）；站名无法识别时返回 {"error": "出发站「…」无法识别，可用 search_station 工具查车站名"}，换成准确站名重试即可。

🇨🇳 中文使用说明

一句话：直接用中文站名查 12306 余票，数据来自 12306 官方公开查询接口，纯只读。

在线直连地址（免费、无需 Key、开箱即用）：https://mcp.pianam.cn/train-mcp/mcp

在 MCP 客户端（Claude Desktop / Cursor / Cherry Studio / Cline 等）的配置里加入：

{  "mcpServers": {    "12306-train": {      "type": "http",      "url": "https://mcp.pianam.cn/train-mcp/mcp"    }  }}

工具：

• query_train_tickets(from_city, to_city, date)：查余票。出发/到达站直接给中文城市或车站名（如「青岛」「北京南」），日期格式 YYYY-MM-DD，留空默认明天；返回车次、出发到达时间、历时及商务座/一等座/二等座/软卧/硬卧/硬座/无座等余票。
• search_station(keyword)：不确定城市有哪些车站时模糊搜索站名（如「北京」返回 北京/北京南/北京西/北京北 等）。
• 仅查票，不能购票；12306 预售期通常 15 天。

服务特性：数据源全部为公开接口、无需注册/付费；服务端内存缓存、失败自动降级/切换备用通道；单 IP 限流 200 次/分钟。

本地部署：

git clone https://github.com/boy-373/12306-train-mcp.gitcd 12306-train-mcppip install -r requirements.txtpython train_mcp_server.py# 默认监听 127.0.0.1:8003，可用环境变量 MCP_HOST / MCP_PORT / MCP_ALLOWED_HOSTS / MCP_ALLOWED_ORIGINS 覆盖

Docker 自托管：

# 方式一：直接拉预构建镜像（最快，无需构建）docker run -d -p 8000:8000 --name 12306-train-mcp ghcr.io/boy-373/12306-train-mcp:latest
# 方式二：从源码构建git clone https://github.com/boy-373/12306-train-mcp.git && cd 12306-train-mcpdocker build -t 12306-train-mcp .docker run -d -p 8000:8000 --name 12306-train-mcp 12306-train-mcp# MCP 地址填：http://127.0.0.1:8000/mcp

容器默认监听 0.0.0.0:8000（可用 -e MCP_PORT=端口 改），无需任何 API Key 或账号。

📄 License

MIT © 2026 boy-373

[12306-train-mcp MCP server]

Install via Smithery

One-click install for Smithery-supported clients (Claude Desktop, Cursor, etc.):

[Smithery]

Or run:

npx -y @smithery/cli install 1561852680/12306-train-mcp --client claude

🔌 Use with Any AI Client — One-Click MCP Gateway

We host a free MCP aggregation gateway: your AI assistant (Cherry Studio, Claude Desktop, Cursor, Cline — any MCP client) discovers and uses MCP tools through a single URL:

https://mcp.pianam.cn/ai/mcp

Add it once, then talk in plain language — the AI automatically searches → inspects → calls the right MCP for you.

• 🔍 Search across 22,000+ indexed MCP servers (Chinese & English)
• ⚡ 50+ live hosted MCPs callable directly out of the box (weather, exchange rates, hot trends, IP geo, train tickets…)
• 📦 GitHub-based MCPs: search finds them with links to install locally
• ✅ Servers health-checked weekly — alive ones ranked first
• 🆓 Free to use, no key required

📖 3-step setup guide / 中文接入教程: https://mcp.pianam.cn/ai-gateway