mcps 0.1.dev12

Multi-Component Platform System (MCPS) - A framework for managing distributed services and agents

MCPS

Multi-Component Platform System (MCPS) - A framework for managing distributed services and agents

MCPS is a comprehensive framework for managing distributed services and intelligent agents. It provides tools for service discovery, lifecycle management, and runtime execution in various environments.

Features

• Agent Discovery: Find and query agents using natural language or capability matching

• Lifecycle Management: Deploy, start, stop, and monitor agents

• Runtime Environments: Support for Python, Docker, and other runtime environments

• Service Integration: Connect agents with tools and services

• Caching Strategy: Efficient caching of messages, logs, and agent state

• CLI Interface: Command-line tools for managing agents and services

Installation

pip install mcps

Quick Start

from mcps import MCPS_Client
from mcps import Agents_Client

# Query for tools
mcps_tools_client = MCPS_Client(api_key="YOUR_KEY")
tools = mcps_tools_client.query("北京的天气怎么样？", topk=1)[0]

# Query for agents
mcps_agent_client = Agents_Client(api_key="YOUR_KEY")
agent = mcps_agent_client.call_from_query("北京的天气怎么样？", topl=1)[0]

# Run agent with tools
result = agent.run(query="北京的天气怎么样？", tools=tools)

CLI Usage

# List available agents
mcps agent list

# Query for agents
mcps agent query "weather forecast"

# Deploy and run an agent
mcps agent deploy weather_agent_1
mcps agent start weather_agent_1
mcps agent run weather_agent_1 "What's the weather in Beijing?"
mcps agent stop weather_agent_1
mcps agent cleanup weather_agent_1

License

MIT License