examples

Examples

This directory contains example scripts demonstrating how to use Agent Control Plane.

Available Examples

Basic Usage (basic_usage.py)

Demonstrates fundamental concepts and basic usage patterns:

• Creating the control plane
• Creating agents with different permission levels
• Executing actions
• Handling permissions and errors

Run:

python examples/basic_usage.py

Advanced Features (advanced_features.py)

Showcases advanced capabilities:

• Mute Agent - Capability-based execution
• Shadow Mode - Simulation without execution
• Constraint Graphs - Multi-dimensional context
• Supervisor Agents - Recursive governance
• Reasoning Telemetry - Tracking agent decisions

Run:

python examples/advanced_features.py

Configuration (configuration.py)

Shows different configuration patterns and agent profiles:

• Development/Testing agent configuration
• Production agent configuration
• Read-only agent configuration
• Multi-tenant configurations

Run:

python examples/configuration.py

Framework Integrations

LangChain Integration (langchain_demo.py)

Demonstrates governance for LangChain agents:

• Basic LangChain adapter setup
• Custom tool mappings for company-specific tools
• Blocked action callbacks for monitoring
• Real-world integration patterns
• Statistics and audit trails

Run:

python examples/langchain_demo.py

MCP Protocol Integration (mcp_demo.py)

Shows how to create governed MCP servers:

• Basic MCP server with governance
• MCP protocol message handling (JSON-RPC)
• Tool and resource registration
• Error handling for blocked actions
• Integration patterns for MCP clients

Run:

python examples/mcp_demo.py

A2A Protocol Integration (a2a_demo.py)

Demonstrates agent-to-agent communication:

• Creating A2A agents with governance
• Agent Cards for discovery
• Task requests and delegation
• Multi-agent coordination
• Secure inter-agent communication

Run:

python examples/a2a_demo.py

OpenAI Adapter (adapter_demo.py)

Complete demonstration of the OpenAI SDK adapter:

• Drop-in middleware for OpenAI client
• Tool call interception and governance
• Custom tool mappings
• Production deployment patterns

Run:

python examples/adapter_demo.py

Creating Your Own Examples

When creating examples:

1. Import from agent_control_plane package
2. Include clear comments explaining each step
3. Use descriptive variable names
4. Show both success and error cases
5. Keep examples focused on specific features

Example template:

"""
Example: Your Feature Name

This example demonstrates how to use [feature name].
"""

from agent_control_plane import AgentControlPlane, create_standard_agent
from agent_control_plane.agent_kernel import ActionType

def example_function():
    """Demonstrates [specific functionality]"""
    # Create control plane
    control_plane = AgentControlPlane()
    
    # Your example code here
    pass

if __name__ == "__main__":
    example_function()

Supported Frameworks and Protocols

The Agent Control Plane supports multiple frameworks and protocols:

• OpenAI SDK: Drop-in adapter for OpenAI client
• LangChain: Governance for LangChain agents and tools
• MCP (Model Context Protocol): Anthropic's standard for tool/resource access
• A2A (Agent-to-Agent): Google/Linux Foundation protocol for agent coordination

All adapters provide the same governance approach with consistent security and audit capabilities.