Skip to main content

What is MCP?

Model Context Protocol (MCP) is a framework that allows external systems to pass structured, real-time information to large language models (LLMs) like GPT during each interaction. It bridges the gap between stateless systems such as REST APIs or command-line tools and the varying contextual needs of modern applications. By supplying this additional context, MCP enables the model to produce more accurate, relevant, and consistent responses. At a technical level, MCP works by attaching a structured JSON payload to the user’s prompt at runtime. This payload carries background details such as the user’s identity, recent actions, task-specific instructions, or relevant domain knowledge. This is referred to as runtime context — data that helps the model understand the situation in which it is being used, without needing to ask for that information in the prompt itself. While invisible in the model’s final response, this context directly influences how the model interprets the prompt and generates output.

What is the Siren MCP server?

When it comes to Siren, the Model Context Protocol (MCP) Server serves as the integration layer that enables LLM-powered environments to interact with the Siren platform using standardized function calls. It acts as a communication bridge, allowing language models to trigger messaging workflows and retrieve delivery insights in real time. Rather than manually composing HTTP requests or managing channel-specific logic, AI agents can call predefined functions like messaging.send, users.add, or workflows.schedule. These are optimized for channels including Email, Slack, SMS, WhatsApp, and others, making it easy to embed messaging actions into intelligent, event-driven workflows. By abstracting the messaging infrastructure behind clean, callable interfaces, the Siren MCP Server allows developers to build smarter, more responsive AI agents that can operate autonomously across user-facing communication channels.

Available Tools in the Siren MCP Server

The Siren MCP server provides a suite of standardized tools that LLM agents can call during runtime. These tools enable seamless integration with Siren’s messaging, template, user, workflow, and webhook infrastructure.

Messaging Tools

Template Management Tools

User Management Tools

Workflow Tools

Webhook Configuration Tools


Tool Categories

To simplify configuration, Siren MCP supports tool categories — predefined groups of related tools that can be referenced using a single flag.

Category Mappings


Getting Started with Siren MCP Server Integrations

To get started with the Siren MCP server, you’ll need an API key to authenticate your access. If you’ve already signed up, you may have received this key during onboarding. Otherwise, you can generate one from the Siren Dashboard under Settings > Configuration > API Keys. Once you have the API key, you can configure the Siren MCP server within any AI application that supports the Model Context Protocol (MCP). While we’ve included integration examples for tools like Clause, VS Code, and CLI-based agents, the same approach applies to any MCP-compatible environment. Learn more about generating and managing API keys →
Add the following to your claude_desktop_config.json. See Claude MCP documentation for more details.
Or if you are using Docker:

Configuration Options

Command Line Arguments

API Key Format

Siren supports multiple API key formats:

Workspace Configuration

If you are working with multiple Siren workspaces, specify the workspace ID:

Environment Variables

You can set environment variables instead of passing API keys as arguments:

Debugging

Debug Mode

Enable verbose logging with the --debug flag:

Using MCP Inspector

The MCP Inspector provides a visual interface for debugging your MCP server.
1

First, build the server:

2

Run with the MCP Inspector:

3

Open the inspector in your browser at `http://localhost:6274`

Open the URL in your web browser to access the debugging interface.

Docker Debugging

For Docker-based debugging:
1

Build the Docker image:

2

Run with MCP Inspector:

Health Check

  • Verify your configuration with a simple health check:
  • If the server starts successfully, you will see:
  • You can also check the version:

Troubleshooting

Some of the common issues and solutions are stated below.
Possible Causes
  • Incorrect format
  • Expired key
  • Key not set properly Solutions
  • Ensure key starts with sk_siren_, sk_test_, or sk_live_
  • Check your environment variables or CLI argument
  • Generate a new API key from the Siren dashboard if needed
Possible Causes
  • Typo in tool name
  • Tool not enabled
  • Outdated MCP package Solutions
  • Verify tool name (case-sensitive)
  • Add the tool to the --tools list in the command
  • Update to the latest MCP version
Possible Causes
  • Network problems
  • Firewall blocking
  • API endpoint down Solutions
  • Check your internet connection
  • Configure firewall exceptions for outbound calls
  • Verify Siren API availability at status.trysiren.io
Possible Causes
  • Insufficient API key permissions
  • Workspace-specific access restrictions Solutions
  • Ensure your API key has required scopes
  • Check your workspace access level
  • Contact Siren Support if the issue persists
Possible Causes
  • Too many requests in a short time
  • Plan quota exceeded Solutions
  • Implement retry logic or request throttling
  • Upgrade your Siren plan if you are consistently hitting limits

Examples

Basic Messaging

Template Management

Workflow Automation

Full Access