Sinores Open
A self-hosted WhatsApp gateway that exposes messaging capabilities via MCP, allowing AI agents to send, receive, and manage WhatsApp messages through a single-command setup with SQLite storage.
README
<div align="center">
sinores-wp-mcp
Give your AI agent access to WhatsApp.
Self-hosted WhatsApp gateway with a built-in MCP server.<br> Connect a number, point your agent at one endpoint, and start building.
Quick start · MCP setup · Hosted Sinores · Contributing
</div>
sinores-wp-mcp is the open-source edition of Sinores. It turns a WhatsApp number into tools that any MCP-compatible AI agent can use. Your agent can discover connected channels, check their status, send messages, and read message history-all through a standard Model Context Protocol endpoint.
No Docker. No Postgres. No Redis. One Node.js process and one port.
Want it managed for you? Sinores Cloud is the hosted, production-ready service. This repository is the free, MIT-licensed core for developers who prefer to run everything themselves.
Why Sinores Open?
- Built for AI agents - MCP is included, not bolted on as an afterthought.
- Start in minutes - install dependencies, add one token, and scan a QR code.
- Keep control of your data - messages, credentials, and sessions stay on your machine.
- Simple by design - the dashboard, API, MCP server, database, and WhatsApp connection run together.
- Works with any MCP client - use it with your favorite agent, IDE, or custom automation.
- Open source - inspect it, extend it, and ship your own integrations under the MIT license.
What you get
| Capability | Description |
|---|---|
| WhatsApp channels | Link numbers using the familiar WhatsApp QR flow |
| Send and receive | Exchange text messages and keep a local history |
| Built-in MCP server | Give agents WhatsApp tools through POST /api/mcp |
| Web dashboard | Create channels, view connection state, scan QR codes, and test messages |
| Local persistence | Store channel data and message history in SQLite |
| Simple authentication | Protect the dashboard, REST API, and MCP endpoint with one admin token |
Quick start
Requirements
- Node.js 22 or newer
- A WhatsApp account that can link a companion device
Run locally
git clone https://github.com/karl4th/sinores-wp-mcp.git
cd sinores-wp-mcp
npm install
cp .env.example .env.local
Generate a secure admin token:
openssl rand -hex 32
Put the result in .env.local:
ADMIN_TOKEN=your-generated-token
Start Sinores Open:
npm run dev
Open http://localhost:3000, sign in with your admin token, create a channel, and scan the QR code from WhatsApp → Linked devices → Link a device.
For a production-style local run:
npm run build
npm start
Connect an AI agent
Sinores exposes a Streamable HTTP MCP endpoint at:
http://localhost:3000/api/mcp
Configure your MCP client with the endpoint and bearer token. The exact config shape depends on the client, but the connection values are:
{
"url": "http://localhost:3000/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_ADMIN_TOKEN"
}
}
Once connected, the agent receives four tools:
| Tool | What it does |
|---|---|
list_channels |
Lists connected WhatsApp channels and their status |
get_channel_status |
Returns the current state of one channel |
send_message |
Sends a text message to a phone number |
list_messages |
Reads message history, with direction and timestamp filters |
Start with list_channels to get a channelId, then pass that ID to the other tools.
How it works
flowchart LR
A["AI agent"] -->|"MCP over HTTP"| S["Sinores Open"]
U["Web dashboard"] --> S
S --> M["WhatsApp via Baileys"]
S --> D[("SQLite")]
- Next.js serves the dashboard, REST routes, and MCP endpoint.
- Baileys manages WhatsApp companion-device sessions.
- SQLite stores channels and message history in
./databy default. - The MCP SDK exposes agent tools from the same application process.
Configuration
| Variable | Required | Default | Purpose |
|---|---|---|---|
ADMIN_TOKEN |
Yes | - | Protects the dashboard, API, and MCP endpoint |
CHANNEL_LIMIT |
No | 10 |
Maximum number of WhatsApp channels |
DATA_DIR |
No | ./data |
SQLite and application data directory |
BAILEYS_AUTH_DIR |
No | ./data/baileys-auth |
WhatsApp session credentials directory |
COOKIE_SECURE |
No | unset | Set to true when serving over HTTPS |
See .env.example for a ready-to-copy configuration file.
Open source or hosted?
| Sinores Open | Sinores Cloud | |
|---|---|---|
| Best for | Developers, prototypes, personal automation | Teams and production workloads |
| Hosting | Your machine or server | Managed for you |
| Database | Local SQLite | Managed infrastructure |
| Authentication | Single admin token | Multi-tenant access controls |
| Operations | You maintain it | Sinores maintains it |
| License | MIT | Commercial service |
Security notes
- Treat
ADMIN_TOKENas a password: anyone who has it can access every connected channel. - Do not commit
.env.local, thedatadirectory, or Baileys session credentials. - Use HTTPS and set
COOKIE_SECURE=truewhen exposing an instance outside your local network. - This project uses Baileys, an unofficial WhatsApp Web API. It is not affiliated with or endorsed by WhatsApp or Meta. Review WhatsApp's terms and use it responsibly.
Contributing
Issues and pull requests are welcome. If you have an idea, found a bug, or built something interesting with Sinores, open an issue.
If Sinores Open helps you build something useful, please give the repository a star. It helps other developers discover the project and motivates us to keep improving it. ⭐
License
Sinores Open is available under the MIT License.
<div align="center">
Built for developers connecting AI agents to the real world.
</div>
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。