Agentor

v1.0.0Coding Agentsstable

Open source version of Claude Managed Agents. Fastest way to build and deploy reliable AI agents, MCP tools and agent-to-agent.

agentsai-agentsai-assistantai-toolsllms
Share:
181
Stars
0
Downloads
0
Weekly
0/5

What is Agentor?

Agentor is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to open source version of claude managed agents. fastest way to build and deploy reliable ai agents, mcp tools and agent-to-agent.

Open source version of Claude Managed Agents. Fastest way to build and deploy reliable AI agents, MCP tools and agent-to-agent.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Open source version of Claude Managed Agents. Fastest way to

Use Cases

Build and deploy reliable AI agents with minimal setup.
Create agent-to-agent communication systems.
Manage MCP tools and agent orchestration.
CelestoAI

Maintainer

LicenseApache-2.0
Languagepython
Versionv1.0.0
UpdatedMay 17, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx agentor

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Agentor

Agentor is an open-source Python framework and MCP server for building and deploying reliable AI agents with multi-agent communication support via the A2A (Agent-to-Agent) Protocol. It provides a minimal, code-first API for defining agents with tools, instructions, and LLM backends, then automatically exposes each agent as a REST API endpoint and as an MCP tool. It is designed for developers who want to build production-grade agent systems without heavyweight frameworks like LangChain or AutoGen, with optional serverless deployment through the Celesto platform.

Prerequisites

  • Python 3.9+ installed
  • pip or uv package manager
  • API key for your chosen LLM provider (e.g. an Anthropic, OpenAI, or Google API key passed as the model parameter)
  • An MCP-compatible client such as Claude Desktop or Claude Code if using the MCP server interface
1

Install the agentor package

Install from PyPI with pip. The celesto CLI is included automatically for deployment workflows.

pip install agentor
2

Define an agent in Python

Create a Python file that instantiates an Agentor agent with a name, model identifier, and optional tools and instructions.

from agentor import Agentor

agent = Agentor(
    name="My Assistant",
    model="anthropic/claude-3.5",
    tools=[],
    instructions="You are a helpful assistant."
)

result = agent.run("Hello, what can you do?")
print(result)
3

Serve the agent as a REST API

Call agent.serve() to start a FastAPI server on localhost:8000. The /chat endpoint accepts POST requests with a message body.

agent.serve()  # starts http://localhost:8000/chat
4

Configure the MCP server

Add agentor to your MCP client configuration so Claude or other assistants can discover and invoke your agents as MCP tools.

5

Add built-in tools

Extend the agent with built-in tools like ShellTool for system commands or GetWeatherTool for weather lookups. Pass them in the tools list at instantiation.

from agentor import Agentor
from agentor.tools import ShellTool

agent = Agentor(
    name="Dev Agent",
    model="anthropic/claude-3.5",
    tools=[ShellTool()]
)

Agentor Examples

Client configuration

Add to your claude_desktop_config.json to register an Agentor-based MCP server. Adjust the script path to your agent file.

{
  "mcpServers": {
    "agentor": {
      "command": "python",
      "args": ["/path/to/your_agent.py"],
      "env": {}
    }
  }
}

Prompts to try

Example interactions with an Agentor-powered agent exposed through MCP.

- "What tools do you have available?"
- "Run the shell command 'ls -la /tmp' and show me the output"
- "What is the weather in London right now?"
- "Start a sub-agent to research a topic and report back to me"
- "List all active agents in this system"

Troubleshooting Agentor

Agent fails with an LLM authentication error

Agentor passes the model string to the underlying LLM provider. Ensure the relevant API key environment variable is set in your shell (e.g. ANTHROPIC_API_KEY, OPENAI_API_KEY, or GOOGLE_API_KEY) before running the agent script.

MCP client cannot connect to the agentor server

Confirm the agent script calls agent.serve_mcp() or is configured to start the MCP transport. Check that the script path in the MCP config args array is an absolute path and the file is executable by Python.

ShellTool or other built-in tools not found on import

Run 'pip install --upgrade agentor' to ensure you have the latest version. The tools module location may differ between versions — check the GitHub README for the current import path.

Frequently Asked Questions about Agentor

What is Agentor?

Agentor is a Model Context Protocol (MCP) server that open source version of claude managed agents. fastest way to build and deploy reliable ai agents, mcp tools and agent-to-agent. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Agentor?

Follow the installation instructions on the Agentor GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Agentor?

Agentor works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Agentor free to use?

Yes, Agentor is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "agentor": { "command": "npx", "args": ["-y", "agentor"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Agentor?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides