# HexaEight MCP Client
🚀 **Framework-agnostic MCP integration with secure LLM configuration, automatic multi-agent coordination, and military-grade encryption.**
Build sophisticated multi-agent systems with minimal code while maintaining complete security and framework choice.
## 🌟 Key Features
- 🔐 **Secure LLM Configuration** - Encrypt API keys and endpoints with HexaEight's military-grade encryption
- 🤖 **Framework Integration** - AutoGen, CrewAI, LangChain with encrypted LLM configs
- 🛠️ **Multi-Agent Types** - LLM agents, Tool agents, User agents with automatic coordination
- 🌐 **Automatic Discovery** - Dynamic capability discovery and intelligent message routing
- 📡 **PubSub Coordination** - Seamless agent communication and task delegation
- ⚡ **One-Line Setup** - From configuration to production-ready agents instantly
- 🎯 **Business Logic** - Write your actual API/database code once in tool agents
- 🔒 **Production Security** - Zero plain-text secrets, secure agent identity management
## 📋 Requirements
### Core Requirements
1. **HexaEight License**
- Purchase from [store.hexaeight.com](https://store.hexaeight.com)
- Install on your development/production machine
- Enables creation of **1 parent configuration** + **unlimited child configurations**
2. **HexaEight Agentic IAM Server** *(Releasing Soon)*
- Create CLIENT Application (CLIENTID)
- Provisions PubSub server bonded to your CLIENTID
- Enables seamless agent communication
3. **Python Environment**
- Python 3.8+
- .NET SDK (for agent creation)
### Agent Licensing Model
| Agent Type | License Requirement | Expiry Behavior |
|------------|--------------------|-----------------|
| **Parent Agents** | ✅ Active HexaEight license required | ❌ Stop working when license expires |
| **Child Agents** | ✅ Created with valid license | ✅ Continue working after license expires |
**Note**: All agents (parent and child) are tied to your CLIENT Application (CLIENTID).
## 🚀 Installation
```bash
# Basic installation
pip install hexaeight-mcp-client
# With framework support
pip install hexaeight-mcp-client[autogen] # For AutoGen
pip install hexaeight-mcp-client[crewai] # For CrewAI
pip install hexaeight-mcp-client[langchain] # For LangChain
pip install hexaeight-mcp-client[all] # All frameworks
```
### Prerequisites
```bash
# Core HexaEight agent package
pip install hexaeight-agent
# Optional: Framework packages
pip install pyautogen # For AutoGen integration
pip install crewai # For CrewAI integration
pip install langchain # For LangChain integration
```
## 🔐 Secure LLM Configuration
### Encrypt Your LLM Configuration
```python
from hexaeight_mcp_client import HexaEightAutoConfig, quick_autogen_llm
# Your sensitive LLM configuration
llm_config = {
"provider": "openai",
"api_key": "sk-your-actual-openai-api-key-here", # Real API key
"model": "gpt-4",
"temperature": 0.7,
"max_tokens": 1000
}
# Create agent for encryption capabilities
agent = await quick_autogen_llm('parent_config.json')
# Encrypt and protect your configuration
protector = HexaEightAutoConfig.create_llm_config_protector(agent.hexaeight_agent)
protector.save_protected_config(llm_config, "secure_openai.enc")
print("✅ API keys encrypted and secured!")
```
### Supported LLM Providers
| Provider | Configuration Example |
|----------|----------------------|
| **OpenAI** | `{"provider": "openai", "api_key": "sk-...", "model": "gpt-4"}` |
| **Azure OpenAI** | `{"provider": "azure_openai", "api_key": "...", "azure_endpoint": "https://..."}` |
| **Anthropic** | `{"provider": "anthropic", "api_key": "sk-ant-...", "model": "claude-3-sonnet"}` |
| **Local Ollama** | `{"provider": "ollama", "base_url": "http://localhost:11434", "model": "llama2"}` |
## 🤖 Framework Integration with Secure LLM Config
### AutoGen Integration
```python
from hexaeight_mcp_client import HexaEightAutoConfig, create_autogen_agent_with_hexaeight
# Create AutoGen agent with encrypted LLM configuration
autogen_agent, hexaeight_agent = await create_autogen_agent_with_hexaeight(
config_file="parent_config.json",
agent_type="parentLLM",
name="IntelligentCoordinator",
system_message="You can coordinate with multiple specialized agents and services."
)
# AutoGen agent now has:
# ✅ Encrypted LLM configuration (your API keys secure)
# ✅ HexaEight coordination tools
# ✅ Automatic capability discovery
# ✅ Multi-agent coordination via PubSub
```
### CrewAI Integration
```python
from hexaeight_mcp_client import create_crewai_agent_with_hexaeight
# Create CrewAI agent with encrypted LLM configuration
crewai_agent, hexaeight_agent = await create_crewai_agent_with_hexaeight(
config_file="parent_config.json",
agent_type="parentLLM",
role="Multi-Agent Coordinator",
goal="Coordinate complex workflows using available tools and services",
backstory="Expert in orchestrating multi-agent systems with secure communication"
)
# CrewAI agent now has:
# ✅ Encrypted LLM configuration
# ✅ Role-based coordination capabilities
# ✅ Access to dynamic tool discovery
# ✅ Secure multi-agent communication
```
### LangChain Integration
```python
from hexaeight_mcp_client import HexaEightAutoConfig
# Create LangChain setup with encrypted LLM configuration
hexaeight_agent = await HexaEightAutoConfig.create_llm_agent_with_protected_config(
agent_type="parentLLM",
config_file="parent_config.json",
encrypted_llm_config_file="secure_openai.enc",
framework="langchain"
)
# Access decrypted LLM config for LangChain
llm_config = hexaeight_agent.llm_config
# Use with LangChain
from langchain.chat_models import ChatOpenAI
from langchain.agents import initialize_agent
llm = ChatOpenAI(
model=llm_config["model"],
openai_api_key=llm_config["api_key"], # Securely decrypted
temperature=llm_config["temperature"]
)
# LangChain agent with HexaEight tools
tools = hexaeight_agent.get_available_tools()
agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION)
```
## 🛠️ Multi-Agent Architecture
### Agent Types
| Agent Type | Purpose | Verification | Broadcasts | Use Case |
|------------|---------|--------------|------------|----------|
| **parentLLM** | LLM coordination | ✅ Required | ✅ Receives all | Main intelligence and coordination |
| **childLLM** | LLM task execution | ✅ Required | ✅ Receives all | Specialized LLM workers |
| **parentTOOL** | Service provision | ❌ Not required | ❌ Ignores broadcasts | API services, databases |
| **childTOOL** | Specialized services | ❌ Not required | ❌ Ignores broadcasts | Specific tool implementations |
| **USER** | Human interaction | ❌ Not required | ❌ Broadcasts to LLMs only | User interfaces |
### Quick Agent Creation
```python
from hexaeight_mcp_client import quick_autogen_llm, quick_tool_agent, quick_user_agent
# LLM Agent (Intelligence Layer)
llm_agent = await quick_autogen_llm(
config_file="parent_config.json",
agent_type="parentLLM"
)
# Tool Agent (Business Logic Layer)
weather_agent = await quick_tool_agent(
config_file="weather_config.json",
service_formats=["weather_request", "forecast_request"],
agent_type="parentTOOL"
)
# User Agent (Interface Layer)
user_agent = await quick_user_agent(
config_file="user_config.json"
)
```
## 🔧 Implementing Business Logic (Tool Agents)
### Weather Service Example
```python
class WeatherService:
async def initialize(self):
# Create tool agent
self.agent = await quick_tool_agent(
config_file="weather_config.json",
service_formats=["weather_request", "location_query"],
agent_type="parentTOOL"
)
# Register your business logic
self.agent.register_service_handler("weather_request", self.handle_weather)
async def handle_weather(self, content, event, tracker):
"""YOUR ACTUAL BUSINESS LOGIC GOES HERE"""
# Parse request
request = json.loads(content)
location = request.get("location", "London")
# 🌍 CALL YOUR ACTUAL WEATHER API
weather_data = await self.call_openweather_api(location)
# Process and enhance data
enhanced_data = self.process_weather_data(weather_data)
# Send structured response
response = {
"type": "weather_response",
"location": location,
"data": enhanced_data,
"processed_by": self.agent.agent_name
}
await self.agent.hexaeight_agent.publish_to_agent(
self.agent.pubsub_url,
"internal_id",
event.sender_internal_id,
json.dumps(response),
"weather_response"
)
return True
async def call_openweather_api(self, location):
"""Your actual API integration"""
# Real API call to OpenWeatherMap, WeatherAPI, etc.
api_key = "your-weather-api-key"
url = f"http://api.openweathermap.org/data/2.5/weather?q={location}&appid={api_key}"
# ... implement actual API call
pass
```
### Database Service Example
```python
class DatabaseService:
async def initialize(self):
self.agent = await quick_tool_agent(
config_file="db_config.json",
service_formats=["database_query", "customer_lookup"],
agent_type="parentTOOL"
)
self.agent.register_service_handler("database_query", self.handle_query)
async def handle_query(self, content, event, tracker):
"""YOUR ACTUAL DATABASE LOGIC"""
# Parse database request
query_data = json.loads(content)
# 🗄️ EXECUTE REAL DATABASE OPERATIONS
results = await self.execute_sql_query(query_data)
# Send results back
response = {
"type": "database_response",
"results": results,
"processed_by": self.agent.agent_name
}
await self.send_response(response, event)
return True
async def execute_sql_query(self, query_data):
"""Your actual database operations"""
# Real database connection and queries
# PostgreSQL, MongoDB, MySQL, etc.
pass
```
## 🌐 Automatic Coordination & Discovery
### Capability Discovery
```python
# LLM agents automatically discover available tools
llm_agent = await quick_autogen_llm("parent_config.json")
# Discover what services are available
capabilities = await llm_agent.capability_discovery.discover_ecosystem_capabilities()
print("Available Services:")
for service_id, info in capabilities["tool_capabilities"].items():
service_name = info["capabilities"]["service_name"]
endpoints = info["capabilities"]["endpoints"]
print(f" 🔧 {service_name}: {endpoints}")
```
### User Interaction
```python
# User requests are automatically routed to appropriate agents
user_agent = await quick_user_agent("user_config.json")
# This broadcast only reaches LLM agents
await user_agent.broadcast_to_llms("Get me the weather in Tokyo and customer data for account 1001")
# LLM agents will:
# 1. Understand the request
# 2. Discover available weather and database services
# 3. Coordinate with appropriate tool agents
# 4. Aggregate results and respond to user
```
## 📡 Complete System Example
```python
import asyncio
from hexaeight_mcp_client import *
async def create_intelligent_system():
"""Create a complete multi-agent system"""
# 1. Setup secure LLM configuration
llm_config = {
"provider": "openai",
"api_key": "sk-your-key",
"model": "gpt-4"
}
agent = await quick_autogen_llm('parent_config.json')
protector = HexaEightAutoConfig.create_llm_config_protector(agent.hexaeight_agent)
protector.save_protected_config(llm_config, "secure_llm.enc")
# 2. Create LLM coordinator with encrypted config
coordinator = await HexaEightAutoConfig.create_llm_agent_with_protected_config(
agent_type="parentLLM",
config_file="parent_config.json",
encrypted_llm_config_file="secure_llm.enc",
framework="autogen"
)
# 3. Create specialized tool agents
weather_service = await quick_tool_agent(
"weather_config.json",
["weather_request", "forecast_request"],
"parentTOOL"
)
database_service = await quick_tool_agent(
"db_config.json",
["database_query", "customer_lookup"],
"parentTOOL"
)
# 4. Create user interface
user_interface = await quick_user_agent("user_config.json")
# 5. Register business logic with tool agents
# (Your actual API calls, database operations, etc.)
# 6. System automatically coordinates!
# - User sends requests to LLM agents
# - LLM agents discover available tools
# - Tool agents process specialized requests
# - Results are coordinated and returned
print("🚀 Intelligent multi-agent system ready!")
# Example user interaction
await user_interface.broadcast_to_llms(
"What's the weather in London and show me premium customer data?"
)
# Run the system
asyncio.run(create_intelligent_system())
```
## 🎯 Use Cases
### Enterprise AI Assistants
- **LLM Agents**: Understand user requests, coordinate responses
- **Tool Agents**: CRM data, financial systems, inventory management
- **User Agents**: Employee interfaces, customer portals
### Data Analysis Platforms
- **LLM Agents**: Interpret analysis requests, generate insights
- **Tool Agents**: Database queries, API data feeds, visualization
- **User Agents**: Analyst interfaces, report generation
### Customer Service Automation
- **LLM Agents**: Natural language understanding, response generation
- **Tool Agents**: Ticketing systems, knowledge bases, payment processing
- **User Agents**: Customer chat interfaces, agent dashboards
### IoT and Smart Systems
- **LLM Agents**: Command interpretation, system orchestration
- **Tool Agents**: Device control, sensor data, automation rules
- **User Agents**: Mobile apps, voice interfaces, dashboards
## 🔒 Security Features
- **🔐 API Key Encryption**: Zero plain-text secrets using HexaEight's military-grade encryption
- **🛡️ Secure Agent Identity**: Cryptographic agent authentication and authorization
- **📡 Encrypted Communication**: All agent messages encrypted via HexaEight PubSub
- **🎯 Message Filtering**: Agents only process relevant messages based on type and format
- **🔍 Capability Isolation**: Tool agents expose only intended service capabilities
- **⚡ Automatic Locking**: Single-attempt message locking prevents processing conflicts
## 📊 Configuration Examples
### Parent Agent Configuration (parent_config.json)
```json
{
"agent_type": "parentLLM",
"client_id": "your-client-id",
"description": "Main LLM coordination agent",
"framework": "autogen"
}
```
### Tool Agent Configuration (weather_config.json)
```json
{
"agent_type": "parentTOOL",
"client_id": "your-client-id",
"service_name": "WeatherAPI",
"description": "Weather data service"
}
```
### Encrypted LLM Configuration (secure_llm.enc)
```
# This file contains encrypted LLM configuration
# Generated by: protector.save_protected_config(llm_config, "secure_llm.enc")
[Encrypted binary data - your API keys are secure!]
```
## 🚀 Getting Started Checklist
- [ ] Purchase HexaEight License from [store.hexaeight.com](https://store.hexaeight.com)
- [ ] Install license on your development machine
- [ ] Create CLIENT Application via HexaEight Agentic IAM Server *(when available)*
- [ ] Install hexaeight-mcp-client: `pip install hexaeight-mcp-client[all]`
- [ ] Create your first LLM configuration and encrypt it
- [ ] Build your first tool agent with actual business logic
- [ ] Test multi-agent coordination
- [ ] Deploy to production with real PubSub server
## 🛠️ Development Tools
```python
from hexaeight_mcp_client import *
# Check package info and framework availability
print_package_info()
# Setup development environment with example configs
setup_development_environment()
# Discover available configuration files
config_files = discover_config_files()
print("Available configs:", config_files)
# Validate agent types and configurations
is_valid = validate_agent_type("parentLLM")
agent_info = get_agent_type_info("parentLLM")
```
## 📚 API Reference
### Core Classes
- `HexaEightMCPClient` - Base MCP client functionality
- `HexaEightLLMAgent` - LLM agents with coordination capabilities
- `HexaEightToolAgent` - Tool agents for business logic
- `HexaEightUserAgent` - User agents for human interaction
- `LLMConfigProtector` - Encrypt/decrypt LLM configurations
### Framework Adapters
- `AutogenAdapter` - Microsoft AutoGen integration
- `CrewAIAdapter` - CrewAI framework integration
- `LangChainAdapter` - LangChain framework integration
- `GenericFrameworkAdapter` - Custom framework integration
### Quick Setup Functions
- `quick_autogen_llm()` - Create AutoGen LLM agent
- `quick_crewai_llm()` - Create CrewAI LLM agent
- `quick_tool_agent()` - Create tool agent
- `quick_user_agent()` - Create user agent
## 🆘 Troubleshooting
### Common Issues
**Q: "Failed to encrypt LLM configuration"**
A: Ensure you have a valid HexaEight agent loaded before creating the config protector.
**Q: "Tool agent not receiving messages"**
A: Verify your service_formats match the message types being sent by LLM agents.
**Q: "LLM verification failed"**
A: Make sure your agent_type is "parentLLM" or "childLLM" and provide appropriate verification responses.
**Q: "Connection to PubSub failed"**
A: Check that your CLIENT Application is properly configured and PubSub server is running.
## 🔗 Links & Resources
- **🏪 HexaEight Store**: [store.hexaeight.com](https://store.hexaeight.com) - Purchase licenses
- **📖 Documentation**: [GitHub Repository](https://github.com/HexaEightTeam/hexaeight-mcp-client)
- **🐛 Issues**: [GitHub Issues](https://github.com/HexaEightTeam/hexaeight-mcp-client/issues)
- **📦 HexaEight Agent**: [hexaeight-agent on PyPI](https://pypi.org/project/hexaeight-agent/)
- **💬 Support**: Contact support for licensing and technical issues
## 📄 License
MIT License - see LICENSE file for details.
---
**🌟 Ready to build intelligent multi-agent systems with enterprise-grade security?**
Start with `pip install hexaeight-mcp-client[all]` and create your first encrypted LLM agent today!
Raw data
{
"_id": null,
"home_page": null,
"name": "hexaeight-mcp-client",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "mcp, agent, hexaeight, multi-agent, identity, security, autogen, crewai, langchain",
"author": null,
"author_email": "HexaEight <support@hexaeight.com>",
"download_url": "https://files.pythonhosted.org/packages/54/a4/09509ae4eb0c828bbc49fa58b6588dbd3de59758ec67e642c6a01a61bf6d/hexaeight_mcp_client-1.6.802.tar.gz",
"platform": null,
"description": "# HexaEight MCP Client\n\n\ud83d\ude80 **Framework-agnostic MCP integration with secure LLM configuration, automatic multi-agent coordination, and military-grade encryption.**\n\nBuild sophisticated multi-agent systems with minimal code while maintaining complete security and framework choice.\n\n## \ud83c\udf1f Key Features\n\n- \ud83d\udd10 **Secure LLM Configuration** - Encrypt API keys and endpoints with HexaEight's military-grade encryption\n- \ud83e\udd16 **Framework Integration** - AutoGen, CrewAI, LangChain with encrypted LLM configs \n- \ud83d\udee0\ufe0f **Multi-Agent Types** - LLM agents, Tool agents, User agents with automatic coordination\n- \ud83c\udf10 **Automatic Discovery** - Dynamic capability discovery and intelligent message routing\n- \ud83d\udce1 **PubSub Coordination** - Seamless agent communication and task delegation\n- \u26a1 **One-Line Setup** - From configuration to production-ready agents instantly\n- \ud83c\udfaf **Business Logic** - Write your actual API/database code once in tool agents\n- \ud83d\udd12 **Production Security** - Zero plain-text secrets, secure agent identity management\n\n## \ud83d\udccb Requirements\n\n### Core Requirements\n\n1. **HexaEight License** \n - Purchase from [store.hexaeight.com](https://store.hexaeight.com)\n - Install on your development/production machine\n - Enables creation of **1 parent configuration** + **unlimited child configurations**\n\n2. **HexaEight Agentic IAM Server** *(Releasing Soon)*\n - Create CLIENT Application (CLIENTID) \n - Provisions PubSub server bonded to your CLIENTID\n - Enables seamless agent communication\n\n3. **Python Environment**\n - Python 3.8+\n - .NET SDK (for agent creation)\n\n### Agent Licensing Model\n\n| Agent Type | License Requirement | Expiry Behavior |\n|------------|--------------------|-----------------| \n| **Parent Agents** | \u2705 Active HexaEight license required | \u274c Stop working when license expires |\n| **Child Agents** | \u2705 Created with valid license | \u2705 Continue working after license expires |\n\n**Note**: All agents (parent and child) are tied to your CLIENT Application (CLIENTID).\n\n## \ud83d\ude80 Installation\n\n```bash\n# Basic installation\npip install hexaeight-mcp-client\n\n# With framework support\npip install hexaeight-mcp-client[autogen] # For AutoGen\npip install hexaeight-mcp-client[crewai] # For CrewAI\npip install hexaeight-mcp-client[langchain] # For LangChain\npip install hexaeight-mcp-client[all] # All frameworks\n```\n\n### Prerequisites\n\n```bash\n# Core HexaEight agent package\npip install hexaeight-agent\n\n# Optional: Framework packages\npip install pyautogen # For AutoGen integration\npip install crewai # For CrewAI integration \npip install langchain # For LangChain integration\n```\n\n## \ud83d\udd10 Secure LLM Configuration\n\n### Encrypt Your LLM Configuration\n\n```python\nfrom hexaeight_mcp_client import HexaEightAutoConfig, quick_autogen_llm\n\n# Your sensitive LLM configuration\nllm_config = {\n \"provider\": \"openai\",\n \"api_key\": \"sk-your-actual-openai-api-key-here\", # Real API key\n \"model\": \"gpt-4\",\n \"temperature\": 0.7,\n \"max_tokens\": 1000\n}\n\n# Create agent for encryption capabilities\nagent = await quick_autogen_llm('parent_config.json')\n\n# Encrypt and protect your configuration\nprotector = HexaEightAutoConfig.create_llm_config_protector(agent.hexaeight_agent)\nprotector.save_protected_config(llm_config, \"secure_openai.enc\")\n\nprint(\"\u2705 API keys encrypted and secured!\")\n```\n\n### Supported LLM Providers\n\n| Provider | Configuration Example |\n|----------|----------------------|\n| **OpenAI** | `{\"provider\": \"openai\", \"api_key\": \"sk-...\", \"model\": \"gpt-4\"}` |\n| **Azure OpenAI** | `{\"provider\": \"azure_openai\", \"api_key\": \"...\", \"azure_endpoint\": \"https://...\"}` |\n| **Anthropic** | `{\"provider\": \"anthropic\", \"api_key\": \"sk-ant-...\", \"model\": \"claude-3-sonnet\"}` |\n| **Local Ollama** | `{\"provider\": \"ollama\", \"base_url\": \"http://localhost:11434\", \"model\": \"llama2\"}` |\n\n## \ud83e\udd16 Framework Integration with Secure LLM Config\n\n### AutoGen Integration\n\n```python\nfrom hexaeight_mcp_client import HexaEightAutoConfig, create_autogen_agent_with_hexaeight\n\n# Create AutoGen agent with encrypted LLM configuration\nautogen_agent, hexaeight_agent = await create_autogen_agent_with_hexaeight(\n config_file=\"parent_config.json\",\n agent_type=\"parentLLM\",\n name=\"IntelligentCoordinator\",\n system_message=\"You can coordinate with multiple specialized agents and services.\"\n)\n\n# AutoGen agent now has:\n# \u2705 Encrypted LLM configuration (your API keys secure)\n# \u2705 HexaEight coordination tools\n# \u2705 Automatic capability discovery\n# \u2705 Multi-agent coordination via PubSub\n```\n\n### CrewAI Integration\n\n```python\nfrom hexaeight_mcp_client import create_crewai_agent_with_hexaeight\n\n# Create CrewAI agent with encrypted LLM configuration\ncrewai_agent, hexaeight_agent = await create_crewai_agent_with_hexaeight(\n config_file=\"parent_config.json\",\n agent_type=\"parentLLM\", \n role=\"Multi-Agent Coordinator\",\n goal=\"Coordinate complex workflows using available tools and services\",\n backstory=\"Expert in orchestrating multi-agent systems with secure communication\"\n)\n\n# CrewAI agent now has:\n# \u2705 Encrypted LLM configuration\n# \u2705 Role-based coordination capabilities\n# \u2705 Access to dynamic tool discovery\n# \u2705 Secure multi-agent communication\n```\n\n### LangChain Integration\n\n```python\nfrom hexaeight_mcp_client import HexaEightAutoConfig\n\n# Create LangChain setup with encrypted LLM configuration\nhexaeight_agent = await HexaEightAutoConfig.create_llm_agent_with_protected_config(\n agent_type=\"parentLLM\",\n config_file=\"parent_config.json\",\n encrypted_llm_config_file=\"secure_openai.enc\",\n framework=\"langchain\"\n)\n\n# Access decrypted LLM config for LangChain\nllm_config = hexaeight_agent.llm_config\n\n# Use with LangChain\nfrom langchain.chat_models import ChatOpenAI\nfrom langchain.agents import initialize_agent\n\nllm = ChatOpenAI(\n model=llm_config[\"model\"],\n openai_api_key=llm_config[\"api_key\"], # Securely decrypted\n temperature=llm_config[\"temperature\"]\n)\n\n# LangChain agent with HexaEight tools\ntools = hexaeight_agent.get_available_tools()\nagent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION)\n```\n\n## \ud83d\udee0\ufe0f Multi-Agent Architecture\n\n### Agent Types\n\n| Agent Type | Purpose | Verification | Broadcasts | Use Case |\n|------------|---------|--------------|------------|----------|\n| **parentLLM** | LLM coordination | \u2705 Required | \u2705 Receives all | Main intelligence and coordination |\n| **childLLM** | LLM task execution | \u2705 Required | \u2705 Receives all | Specialized LLM workers |\n| **parentTOOL** | Service provision | \u274c Not required | \u274c Ignores broadcasts | API services, databases |\n| **childTOOL** | Specialized services | \u274c Not required | \u274c Ignores broadcasts | Specific tool implementations |\n| **USER** | Human interaction | \u274c Not required | \u274c Broadcasts to LLMs only | User interfaces |\n\n### Quick Agent Creation\n\n```python\nfrom hexaeight_mcp_client import quick_autogen_llm, quick_tool_agent, quick_user_agent\n\n# LLM Agent (Intelligence Layer)\nllm_agent = await quick_autogen_llm(\n config_file=\"parent_config.json\",\n agent_type=\"parentLLM\"\n)\n\n# Tool Agent (Business Logic Layer) \nweather_agent = await quick_tool_agent(\n config_file=\"weather_config.json\",\n service_formats=[\"weather_request\", \"forecast_request\"],\n agent_type=\"parentTOOL\"\n)\n\n# User Agent (Interface Layer)\nuser_agent = await quick_user_agent(\n config_file=\"user_config.json\"\n)\n```\n\n## \ud83d\udd27 Implementing Business Logic (Tool Agents)\n\n### Weather Service Example\n\n```python\nclass WeatherService:\n async def initialize(self):\n # Create tool agent\n self.agent = await quick_tool_agent(\n config_file=\"weather_config.json\",\n service_formats=[\"weather_request\", \"location_query\"],\n agent_type=\"parentTOOL\"\n )\n \n # Register your business logic\n self.agent.register_service_handler(\"weather_request\", self.handle_weather)\n \n async def handle_weather(self, content, event, tracker):\n \"\"\"YOUR ACTUAL BUSINESS LOGIC GOES HERE\"\"\"\n \n # Parse request\n request = json.loads(content)\n location = request.get(\"location\", \"London\")\n \n # \ud83c\udf0d CALL YOUR ACTUAL WEATHER API\n weather_data = await self.call_openweather_api(location)\n \n # Process and enhance data\n enhanced_data = self.process_weather_data(weather_data)\n \n # Send structured response\n response = {\n \"type\": \"weather_response\",\n \"location\": location,\n \"data\": enhanced_data,\n \"processed_by\": self.agent.agent_name\n }\n \n await self.agent.hexaeight_agent.publish_to_agent(\n self.agent.pubsub_url,\n \"internal_id\",\n event.sender_internal_id,\n json.dumps(response),\n \"weather_response\"\n )\n \n return True\n \n async def call_openweather_api(self, location):\n \"\"\"Your actual API integration\"\"\"\n # Real API call to OpenWeatherMap, WeatherAPI, etc.\n api_key = \"your-weather-api-key\"\n url = f\"http://api.openweathermap.org/data/2.5/weather?q={location}&appid={api_key}\"\n # ... implement actual API call\n pass\n```\n\n### Database Service Example\n\n```python\nclass DatabaseService:\n async def initialize(self):\n self.agent = await quick_tool_agent(\n config_file=\"db_config.json\",\n service_formats=[\"database_query\", \"customer_lookup\"],\n agent_type=\"parentTOOL\"\n )\n \n self.agent.register_service_handler(\"database_query\", self.handle_query)\n \n async def handle_query(self, content, event, tracker):\n \"\"\"YOUR ACTUAL DATABASE LOGIC\"\"\"\n \n # Parse database request\n query_data = json.loads(content)\n \n # \ud83d\uddc4\ufe0f EXECUTE REAL DATABASE OPERATIONS\n results = await self.execute_sql_query(query_data)\n \n # Send results back\n response = {\n \"type\": \"database_response\",\n \"results\": results,\n \"processed_by\": self.agent.agent_name\n }\n \n await self.send_response(response, event)\n return True\n \n async def execute_sql_query(self, query_data):\n \"\"\"Your actual database operations\"\"\"\n # Real database connection and queries\n # PostgreSQL, MongoDB, MySQL, etc.\n pass\n```\n\n## \ud83c\udf10 Automatic Coordination & Discovery\n\n### Capability Discovery\n\n```python\n# LLM agents automatically discover available tools\nllm_agent = await quick_autogen_llm(\"parent_config.json\")\n\n# Discover what services are available\ncapabilities = await llm_agent.capability_discovery.discover_ecosystem_capabilities()\n\nprint(\"Available Services:\")\nfor service_id, info in capabilities[\"tool_capabilities\"].items():\n service_name = info[\"capabilities\"][\"service_name\"]\n endpoints = info[\"capabilities\"][\"endpoints\"]\n print(f\" \ud83d\udd27 {service_name}: {endpoints}\")\n```\n\n### User Interaction\n\n```python\n# User requests are automatically routed to appropriate agents\nuser_agent = await quick_user_agent(\"user_config.json\")\n\n# This broadcast only reaches LLM agents\nawait user_agent.broadcast_to_llms(\"Get me the weather in Tokyo and customer data for account 1001\")\n\n# LLM agents will:\n# 1. Understand the request\n# 2. Discover available weather and database services \n# 3. Coordinate with appropriate tool agents\n# 4. Aggregate results and respond to user\n```\n\n## \ud83d\udce1 Complete System Example\n\n```python\nimport asyncio\nfrom hexaeight_mcp_client import *\n\nasync def create_intelligent_system():\n \"\"\"Create a complete multi-agent system\"\"\"\n \n # 1. Setup secure LLM configuration\n llm_config = {\n \"provider\": \"openai\",\n \"api_key\": \"sk-your-key\",\n \"model\": \"gpt-4\"\n }\n \n agent = await quick_autogen_llm('parent_config.json')\n protector = HexaEightAutoConfig.create_llm_config_protector(agent.hexaeight_agent)\n protector.save_protected_config(llm_config, \"secure_llm.enc\")\n \n # 2. Create LLM coordinator with encrypted config\n coordinator = await HexaEightAutoConfig.create_llm_agent_with_protected_config(\n agent_type=\"parentLLM\",\n config_file=\"parent_config.json\",\n encrypted_llm_config_file=\"secure_llm.enc\",\n framework=\"autogen\"\n )\n \n # 3. Create specialized tool agents\n weather_service = await quick_tool_agent(\n \"weather_config.json\",\n [\"weather_request\", \"forecast_request\"],\n \"parentTOOL\"\n )\n \n database_service = await quick_tool_agent(\n \"db_config.json\", \n [\"database_query\", \"customer_lookup\"],\n \"parentTOOL\"\n )\n \n # 4. Create user interface\n user_interface = await quick_user_agent(\"user_config.json\")\n \n # 5. Register business logic with tool agents\n # (Your actual API calls, database operations, etc.)\n \n # 6. System automatically coordinates!\n # - User sends requests to LLM agents\n # - LLM agents discover available tools\n # - Tool agents process specialized requests\n # - Results are coordinated and returned\n \n print(\"\ud83d\ude80 Intelligent multi-agent system ready!\")\n \n # Example user interaction\n await user_interface.broadcast_to_llms(\n \"What's the weather in London and show me premium customer data?\"\n )\n\n# Run the system\nasyncio.run(create_intelligent_system())\n```\n\n## \ud83c\udfaf Use Cases\n\n### Enterprise AI Assistants\n- **LLM Agents**: Understand user requests, coordinate responses\n- **Tool Agents**: CRM data, financial systems, inventory management\n- **User Agents**: Employee interfaces, customer portals\n\n### Data Analysis Platforms \n- **LLM Agents**: Interpret analysis requests, generate insights\n- **Tool Agents**: Database queries, API data feeds, visualization\n- **User Agents**: Analyst interfaces, report generation\n\n### Customer Service Automation\n- **LLM Agents**: Natural language understanding, response generation \n- **Tool Agents**: Ticketing systems, knowledge bases, payment processing\n- **User Agents**: Customer chat interfaces, agent dashboards\n\n### IoT and Smart Systems\n- **LLM Agents**: Command interpretation, system orchestration\n- **Tool Agents**: Device control, sensor data, automation rules\n- **User Agents**: Mobile apps, voice interfaces, dashboards\n\n## \ud83d\udd12 Security Features\n\n- **\ud83d\udd10 API Key Encryption**: Zero plain-text secrets using HexaEight's military-grade encryption\n- **\ud83d\udee1\ufe0f Secure Agent Identity**: Cryptographic agent authentication and authorization \n- **\ud83d\udce1 Encrypted Communication**: All agent messages encrypted via HexaEight PubSub\n- **\ud83c\udfaf Message Filtering**: Agents only process relevant messages based on type and format\n- **\ud83d\udd0d Capability Isolation**: Tool agents expose only intended service capabilities\n- **\u26a1 Automatic Locking**: Single-attempt message locking prevents processing conflicts\n\n## \ud83d\udcca Configuration Examples\n\n### Parent Agent Configuration (parent_config.json)\n```json\n{\n \"agent_type\": \"parentLLM\",\n \"client_id\": \"your-client-id\",\n \"description\": \"Main LLM coordination agent\",\n \"framework\": \"autogen\"\n}\n```\n\n### Tool Agent Configuration (weather_config.json)\n```json\n{\n \"agent_type\": \"parentTOOL\",\n \"client_id\": \"your-client-id\", \n \"service_name\": \"WeatherAPI\",\n \"description\": \"Weather data service\"\n}\n```\n\n### Encrypted LLM Configuration (secure_llm.enc)\n```\n# This file contains encrypted LLM configuration\n# Generated by: protector.save_protected_config(llm_config, \"secure_llm.enc\")\n[Encrypted binary data - your API keys are secure!]\n```\n\n## \ud83d\ude80 Getting Started Checklist\n\n- [ ] Purchase HexaEight License from [store.hexaeight.com](https://store.hexaeight.com)\n- [ ] Install license on your development machine\n- [ ] Create CLIENT Application via HexaEight Agentic IAM Server *(when available)*\n- [ ] Install hexaeight-mcp-client: `pip install hexaeight-mcp-client[all]`\n- [ ] Create your first LLM configuration and encrypt it\n- [ ] Build your first tool agent with actual business logic\n- [ ] Test multi-agent coordination\n- [ ] Deploy to production with real PubSub server\n\n## \ud83d\udee0\ufe0f Development Tools\n\n```python\nfrom hexaeight_mcp_client import *\n\n# Check package info and framework availability\nprint_package_info()\n\n# Setup development environment with example configs\nsetup_development_environment()\n\n# Discover available configuration files\nconfig_files = discover_config_files()\nprint(\"Available configs:\", config_files)\n\n# Validate agent types and configurations \nis_valid = validate_agent_type(\"parentLLM\")\nagent_info = get_agent_type_info(\"parentLLM\")\n```\n\n## \ud83d\udcda API Reference\n\n### Core Classes\n- `HexaEightMCPClient` - Base MCP client functionality\n- `HexaEightLLMAgent` - LLM agents with coordination capabilities\n- `HexaEightToolAgent` - Tool agents for business logic\n- `HexaEightUserAgent` - User agents for human interaction\n- `LLMConfigProtector` - Encrypt/decrypt LLM configurations\n\n### Framework Adapters\n- `AutogenAdapter` - Microsoft AutoGen integration\n- `CrewAIAdapter` - CrewAI framework integration \n- `LangChainAdapter` - LangChain framework integration\n- `GenericFrameworkAdapter` - Custom framework integration\n\n### Quick Setup Functions\n- `quick_autogen_llm()` - Create AutoGen LLM agent\n- `quick_crewai_llm()` - Create CrewAI LLM agent\n- `quick_tool_agent()` - Create tool agent\n- `quick_user_agent()` - Create user agent\n\n## \ud83c\udd98 Troubleshooting\n\n### Common Issues\n\n**Q: \"Failed to encrypt LLM configuration\"** \nA: Ensure you have a valid HexaEight agent loaded before creating the config protector.\n\n**Q: \"Tool agent not receiving messages\"** \nA: Verify your service_formats match the message types being sent by LLM agents.\n\n**Q: \"LLM verification failed\"** \nA: Make sure your agent_type is \"parentLLM\" or \"childLLM\" and provide appropriate verification responses.\n\n**Q: \"Connection to PubSub failed\"** \nA: Check that your CLIENT Application is properly configured and PubSub server is running.\n\n## \ud83d\udd17 Links & Resources\n\n- **\ud83c\udfea HexaEight Store**: [store.hexaeight.com](https://store.hexaeight.com) - Purchase licenses\n- **\ud83d\udcd6 Documentation**: [GitHub Repository](https://github.com/HexaEightTeam/hexaeight-mcp-client)\n- **\ud83d\udc1b Issues**: [GitHub Issues](https://github.com/HexaEightTeam/hexaeight-mcp-client/issues)\n- **\ud83d\udce6 HexaEight Agent**: [hexaeight-agent on PyPI](https://pypi.org/project/hexaeight-agent/)\n- **\ud83d\udcac Support**: Contact support for licensing and technical issues\n\n## \ud83d\udcc4 License\n\nMIT License - see LICENSE file for details.\n\n---\n\n**\ud83c\udf1f Ready to build intelligent multi-agent systems with enterprise-grade security?**\n\nStart with `pip install hexaeight-mcp-client[all]` and create your first encrypted LLM agent today!\n",
"bugtrack_url": null,
"license": null,
"summary": "Framework-agnostic MCP client integration for HexaEight agents",
"version": "1.6.802",
"project_urls": {
"Documentation": "https://github.com/HexaEightTeam/hexaeight-mcp-client/blob/main/README.md",
"Homepage": "https://github.com/HexaEightTeam/hexaeight-mcp-client",
"Issues": "https://github.com/HexaEightTeam/hexaeight-mcp-client/issues",
"Repository": "https://github.com/HexaEightTeam/hexaeight-mcp-client.git"
},
"split_keywords": [
"mcp",
" agent",
" hexaeight",
" multi-agent",
" identity",
" security",
" autogen",
" crewai",
" langchain"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "e4cc9ac6df9c7b6d0b63ccb27a733fbb19cd154c03adc04ba84812e31d8cb953",
"md5": "5065e2aa372a3c45f4920adc7f536f5b",
"sha256": "910fed3415d5c1a2013df31f95453b860efb3bf5eeaadbb99a65625b10e6789a"
},
"downloads": -1,
"filename": "hexaeight_mcp_client-1.6.802-py3-none-any.whl",
"has_sig": false,
"md5_digest": "5065e2aa372a3c45f4920adc7f536f5b",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 30616,
"upload_time": "2025-07-15T10:51:39",
"upload_time_iso_8601": "2025-07-15T10:51:39.243711Z",
"url": "https://files.pythonhosted.org/packages/e4/cc/9ac6df9c7b6d0b63ccb27a733fbb19cd154c03adc04ba84812e31d8cb953/hexaeight_mcp_client-1.6.802-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "54a409509ae4eb0c828bbc49fa58b6588dbd3de59758ec67e642c6a01a61bf6d",
"md5": "82245c87acf0732a02044193afd2cc13",
"sha256": "ef3b83bed458471d3afe99306bdd044eeb7ced5e6a5418bb1e956aac91af7cd4"
},
"downloads": -1,
"filename": "hexaeight_mcp_client-1.6.802.tar.gz",
"has_sig": false,
"md5_digest": "82245c87acf0732a02044193afd2cc13",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 33967,
"upload_time": "2025-07-15T10:51:40",
"upload_time_iso_8601": "2025-07-15T10:51:40.030374Z",
"url": "https://files.pythonhosted.org/packages/54/a4/09509ae4eb0c828bbc49fa58b6588dbd3de59758ec67e642c6a01a61bf6d/hexaeight_mcp_client-1.6.802.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-07-15 10:51:40",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "HexaEightTeam",
"github_project": "hexaeight-mcp-client",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "hexaeight-mcp-client"
}
JSON
