Parallel-Poke MCP
Bridges Parallel's Monitor API with Poke to receive SMS notifications when web monitors detect changes and conversationally create/manage monitors through text messages.
README
Parallel-Poke MCP
A bridge between Parallel's Monitor API and Poke - get SMS notifications when your web monitors detect changes, and manage monitors conversationally through Poke.
What It Does
- Receives webhooks from Parallel when your monitors detect web changes
- Forwards event details to Poke which sends you an SMS notification
- Exposes MCP tools so Poke can create/manage monitors on your behalf
This means you can text Poke things like:
- "Create a monitor for React changelog updates, check daily"
- "What monitors do I have?"
- "Show me recent events"
- "Delete the OpenAI monitor"
And you'll automatically get SMS notifications when any monitor detects changes.
Deploy
Option 1: Render (Recommended)
Option 2: Railway
Option 3: Manual Deployment
- Fork this repository
- Deploy to any platform that supports Python (Render, Railway, Fly.io, etc.)
- Set the environment variables (see below)
- Your server will be available at
https://your-app-url.com/mcp
Configuration
Set these environment variables in your deployment platform:
| Variable | Required | Description |
|---|---|---|
PARALLEL_API_KEY |
Yes | Your Parallel API key from platform.parallel.ai |
POKE_API_KEY |
Yes | Your Poke API key from poke.com/settings/advanced |
WEBHOOK_BASE_URL |
Yes | Your deployed server URL (e.g., https://your-app.onrender.com) |
PARALLEL_WEBHOOK_SECRET |
No | Webhook signing secret for verification (recommended) |
MCP_AUTH_TOKEN |
No | If set, MCP clients must send Authorization: Bearer <token> |
NOTIFY_COMPLETION_EVENTS |
No | If true, send completion (no-change) notifications (default false) |
Connect to Poke
- Deploy the server and note your URL (e.g.,
https://parallel-poke-mcp.onrender.com) - Go to poke.com/settings/connections
- Add a new MCP integration with URL:
https://your-url.com/mcp - Test by texting Poke: "List my Parallel monitors"
Endpoints
| Endpoint | Purpose |
|---|---|
/mcp |
MCP protocol endpoint (Poke connects here) |
/webhook/parallel |
Receives webhooks from Parallel monitors |
/health |
Health check for load balancers |
MCP Auth (Optional)
If you set MCP_AUTH_TOKEN, the MCP endpoint requires a bearer token:
Authorization: Bearer <your token>
If your MCP client supports auth (e.g., Poke), configure the same token there.
MCP Tools
The server exposes these tools to Poke:
| Tool | Description |
|---|---|
create_monitor |
Create a new monitor with a query and cadence (hourly/daily/weekly) |
list_monitors |
List all your active monitors |
get_monitor |
Get details of a specific monitor |
update_monitor |
Update a monitor (cadence, webhook, metadata) |
delete_monitor |
Delete a monitor |
list_recent_events |
Get recent events across monitors |
Webhook event types supported by the API:
monitor.event.detectedmonitor.execution.completedmonitor.execution.failed
Use update_monitor to set webhook_event_types if you want completion/failure notifications.
By default, completion events are ignored. Set NOTIFY_COMPLETION_EVENTS=true to send them.
Metadata values must be strings (Parallel API requirement).
Local Development
# Clone the repo
git clone https://github.com/YOUR_USERNAME/parallel-poke-mcp.git
cd parallel-poke-mcp
# Create virtual environment
python3 -m venv venv
source venv/bin/activate # or `venv\Scripts\activate` on Windows
# Install dependencies
python3 -m pip install -r requirements.txt
# Set environment variables
cp .env.example .env
# Edit .env with your API keys
# Run the server
python3 src/server.py
Test with the MCP Inspector:
npx @modelcontextprotocol/inspector
Connect to http://localhost:8000/mcp using "Streamable HTTP" transport.
How It Works
┌─────────────────────┐
│ Parallel Monitors │ You define what to watch
│ (hourly/daily/wk) │ "React changelog updates"
└─────────┬───────────┘
│ webhook fires when changes detected
▼
┌─────────────────────┐
│ This MCP Server │ Receives webhook, fetches
│ (Render/Railway) │ full event details
└─────────┬───────────┘
│ forwards to Poke API
▼
┌─────────────────────┐
│ Poke │ AI formats and sends
└─────────┬───────────┘ SMS notification
│
▼
[SMS]
Plus, Poke can call the MCP tools to manage monitors:
You: "Create a monitor for Vercel changelog, check weekly"
│
▼
┌─────────────────────┐
│ Poke │ Calls create_monitor tool
└─────────┬───────────┘
│ MCP tool call
▼
┌─────────────────────┐
│ This MCP Server │ Calls Parallel API
└─────────┬───────────┘ to create monitor
│
▼
┌─────────────────────┐
│ Parallel API │ Monitor created!
└─────────────────────┘
Example Use Cases
- Changelog monitoring: Track updates to tools you use (React, Next.js, etc.)
- News alerts: Monitor announcements from companies you follow
- Price tracking: Watch for deals on products
- Competitor monitoring: Track competitor announcements or launches
- Job postings: Monitor career pages at companies you're interested in
Troubleshooting
Poke isn't calling my MCP tools
- Make sure the MCP URL ends with
/mcp(e.g.,https://your-app.onrender.com/mcp) - Try sending
clearhistoryto Poke to reset conversation state - Ask Poke explicitly: "Use the Parallel Monitor integration to list monitors"
Webhooks aren't working
- Check that
WEBHOOK_BASE_URLis set correctly (no trailing slash) - Verify your Parallel webhook secret matches if you're using signature verification
- Check the
/healthendpoint to confirm the server is running
Monitor creation fails
- Ensure
PARALLEL_API_KEYis set correctly - Check that
WEBHOOK_BASE_URLis accessible from the internet
License
MIT
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。