Swarms API MCP Server

What is Swarms API MCP Server?

It is an MCP (Model Context Protocol) server that wraps the Swarms REST API, providing tools to run swarm completions and list available swarm types. It is aimed at developers who want to orchestrate multi-agent AI systems using any MCP-compatible client.

How to use Swarms API MCP Server?

Set the SWARMS_API_KEY environment variable with your API key. Run the server (e.g., via SSE transport). The tool swarm_completion accepts a SwarmSpec object and returns the execution result; swarms_available lists supported swarm architectures.

Key features of Swarms API MCP Server

• Provides swarm_completion and swarms_available tools.
• Supports multiple swarm architectures (Concurrent, Sequential, Hybrid, etc.).
• Flexible model support (GPT-4o, Claude, Deepseek, custom models).
• Supabase integration for logging, API key management, and authentication.
• Real-time monitoring, batch processing, and job scheduling.
• Usage tracking with credit consumption.

Use cases of Swarms API MCP Server

• Run multi-agent analysis tasks (e.g., market or financial analysis).
• Schedule recurring swarm executions for automated workflows.
• Batch process multiple swarm tasks simultaneously for high throughput.
• Orchestrate complex agent pipelines from any MCP-compatible interface.

FAQ from Swarms API MCP Server

What dependencies are required?

Python, FastMCP, requests, httpx, pydantic, swarms, and python-dotenv. The server uses FastMCP as the framework.

How is authentication handled?

All endpoints require an API key passed in the x-api-key HTTP header. The server reads the key from the SWARMS_API_KEY environment variable.

Where does data live?

The MCP server communicates with the Swarms cloud API (https://swarms-api-285321057562.us-east1.run.app). Input and output data are processed on Swarms servers.

What transport does it use?

The example server runs with SSE (Server-Sent Events) transport ( mcp.run(transport="sse") ). The REST API also supports standard HTTP.

Are there known limitations?

Usage consumes credits based on number of agents, token count, model selection, and time of day. An API key and sufficient credits are required. Rate limits apply (429 on excess).