home / mcp / lw mcp agents mcp server

LW MCP Agents MCP Server

Lightweight simple MCP agents setup

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "autumn-ais-lw-mcp-agents": {
      "command": "/bin/bash",
      "args": [
        "-c",
        "/path/to/venv/bin/python /path/to/agent_runner.py --config=/path/to/agent_config.json --server-mode"
      ],
      "env": {
        "PATH": "/path/to/venv/bin:/usr/local/bin:/usr/bin",
        "PYTHONPATH": "/path/to/project"
      }
    }
  }
}

LW MCP Agents lets you design and run multi-agent systems using simple configuration files. Each agent can specialize, collaborate, delegate, and reason, all without writing orchestration code. It integrates with MCP-native platforms and supports scalable, recursive reasoning through configurable agents and tools.

How to use

You use an MCP client to connect to the agent servers you configure. Start by running the agent runners described below, then initiate a multi-agent setup where one agent orchestrates others to tackle complex tasks. You’ll be able to deploy research agents, orchestrators, and domain-specific helpers that can scale reasoning and share capabilities across agents.

How to install

Prerequisites you need before installation include Python and a virtual environment manager. You will clone a project, install dependencies, and run the agents with their configurations.

git clone https://github.com/Autumn-AIs/LW-MCP-agents.git
cd LW-MCP-agents
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

Additional setup notes

Two example configurations show how to run agents in a local, MCP-enabled setup. One config runs a research agent as a server-side component and another runs a master orchestrator that delegates tasks to the research agent.

Claude Desktop Integration

You can configure agents to run inside Claude Desktop by editing the Claude MCP configuration to include your agent as an MCP server. The snippet below shows how to reference the agent runner so it starts in server mode with the appropriate environment.

json
{
  "mcpServers": {
    "research_agent": {
      "command": "/bin/bash",
      "args": ["-c", "/path/to/venv/bin/python /path/to/agent_runner.py --config=/path/to/agent_config.json --server-mode"],
      "env": {
        "PYTHONPATH": "/path/to/project",
        "PATH": "/path/to/venv/bin:/usr/local/bin:/usr/bin"
      }
    }
  }
}

Available tools

research_topic

Orchestrator capability to initiate and manage a research task, routing results between agents and tools.

brave_web_search

Tool used by agents to perform web searches as part of the research capability.