Agentify

3. Environment Variable Injection in MCP Configuration

Claude Desktop'sclaude_desktop_config.json:

{ "mcpServers": { "agentify": { "command": "agentify-mcp", "env": { "AGENTIFY_WEBHOOK_URL": "https://webhook.site/your-unique-id", "LOG_LEVEL": "info" } } } }

- macOS:~/Library/Application Support/Claude/claude_desktop_config.json
- Windows:%APPDATA%/Claude/claude_desktop_config.json
- Linux:~/.config/Claude/claude_desktop_config.json

AI will automatically call tools like this:

// When starting a task task_started({ taskDescription: 'Starting React component refactoring', }); // When completing a task task_completed({ taskDescription: 'Completed React component refactoring', outcome: 'success', details: '20% performance improvement', });

When the server starts, you can see the status:

🚀 Agentify MCP Server 📋 Webhook: ✅ Enabled # When URL is configured 📝 Log Level: info

📋 Webhook: ❌ Disabled # When URL is not configured

All tool calls send webhooks in this format:

{ "timestamp": "2024-01-01T12:00:00.000Z", "event": "tool_called", "toolName": "task-started", "arguments": { "taskDescription": "Starting React component refactoring" } }

{ "timestamp": "2024-01-01T12:00:00.000Z", "event": "tool_completed", "toolName": "task-completed", "arguments": { "taskDescription": "Completed React component refactoring", "outcome": "success" }, "result": { "content": ](https://webhook.site)[{ "type": "text", "text": "✅ Task Completed..." }] }, "duration": 150 }

import { AgentifyMCPServer } from 'agentify-mcp'; const server = new AgentifyMCPServer({ webhookUrl: 'https://webhook.site/your-unique-id', logLevel: 'info', }); await server.start();

const server = new AgentifyMCPServer(); // Set webhook URL later server.setWebhookUrl('https://webhook.site/your-unique-id'); // Check webhook status console.log(server.isWebhookEnabled()); // true/false

git clone https://github.com/agentify/agentify-mcp.git cd agentify-mcp npm install # Run in development mode npm run dev # Build npm run build # Run tests npm test # Type checking npm run typecheck # Linting npm run lint

- Completely restart Claude Desktop after environment variable changes

- Check if requests are received in real-time
- Check configuration file path
- Verify JSON syntax
- Confirm command path(agentify-mcpornpx agentify-mcp)

- Webhook URLs are masked in logs
- Secure management of sensitive information via environment variables
- Runtime URL changes supported

# Run without webhook agentify-mcp # Run with webhook AGENTIFY_WEBHOOK_URL="https://webhook.site/abc123" agentify-mcp

- Generate URL from Webhook.site
- Add configuration toclaude_desktop_config.json
- Restart Claude Desktop
- Receive real-time notifications whenever AI performs tasks

# Run all tests npm test # Run tests in watch mode npm run test:watch # Generate coverage report npm run test:coverage

interface ServerConfig { webhookUrl?: string; logLevel?: 'debug' | 'info' | 'warn' | 'error'; }

- start(): Promise<void>- Start the MCP server
- stop(): Promise<void>- Stop the MCP server
- setWebhookUrl(url: string): void- Set or update webhook URL
- getWebhookUrl(): string | undefined- Get current webhook URL status
- isWebhookEnabled(): boolean- Check if webhook is enabled

- Description: Call when starting any task or work
- Parameters:

- taskDescription(string): Brief description of what was started

- Description: Automatically monitors long-running task progress
- Parameters:

- taskThresholdSeconds(number, optional): Auto-trigger threshold in seconds (default: 30)

- Description: Call when finishing any task or work
- Parameters:

- taskDescription(string): Brief description of what was completed
- outcome('success' | 'partial' | 'failed'): Task completion outcome
- details(string, optional): Additional completion details

MIT License - see theLICENSEfile for details.

A multi-agent mesh network designed for completing AI tasks in parallel.