Home / MCP / Near-Intents MCP Agentkit Server

Near-Intents MCP Agentkit Server

Provides an MCP server for AI agents and task orchestration using the CrewAI framework.

python

6stars

View on GitHub

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
    "mcpServers": {
        "crew_ai": {
            "command": "python3",
            "args": [
                "src/crew_server.py"
            ],
            "env": {
                "OPENAI_API_KEY": "YOUR_API_KEY"
            }
        }
    }
}

This MCP server enables AI agent creation, task management, and coordinated workflows using the CrewAI framework. It lets you define agents, assign tasks, and run crews to automate complex AI-driven processes, all through a configurable, local or remote server.

How to use

You interact with the Crew AI MCP server through a client that can send tool requests and orchestrate agents and tasks. Start by creating an agent that defines a role and goal, then generate a task for that agent, and finally assemble a crew to execute the task chain. You can run the crew to see the end-to-end workflow in action, with verbose output to understand each step of the process.

How to install

Prerequisites you need before installation include Python 3.8 or higher, the jq command-line tool for the setup workflow, and a suitable shell environment.

Follow these concrete steps to install and prepare the MCP server:

git clone <your-fork-or-clone-method>

# or use your preferred method to obtain the project

cd <repository-directory>

# Run the setup script to install dependencies and configure MCP settings
./crew.sh

Additional configuration and notes

Configure your OpenAI API key in the environment before running the server. The key provides access to OpenAI services used by agents and tasks.

export OPENAI_API_KEY="your-api-key"

Available tools

create_agent

Create a new AI agent with a specified role, goal, and backstory to guide its behavior.

create_task

Create a task describing the work to be performed by a specified agent, including expected outputs.

create_crew

Assemble agents and tasks into a crew and run the workflow with optional verbose output to monitor progress.