MCP Compatibility Overview

IntentusNet provides an adapter layer for the Model Context Protocol (MCP), enabling MCP-compatible tools to benefit from IntentusNet's execution guarantees.

What is MCP?

The Model Context Protocol is a standardized protocol for:

• Tool definitions and discovery
• Context passing between models and tools
• Structured message formats for AI applications

IntentusNet + MCP

IntentusNet complements MCP by adding:

MCP Provides	IntentusNet Adds
Tool call format	Deterministic routing
Context passing	Execution recording
Discovery	Crash recovery
Message schema	Replay capability

Architecture

The MCPAdapter

IntentusNet provides MCPAdapter to bridge MCP and IntentusNet:

from intentusnet import IntentusClient
from intentusnet.transport import InProcessTransport
from intentusnet.transport.mcp import MCPAdapter

# Create IntentusNet runtime
runtime = IntentusRuntime(enable_recording=True)
transport = InProcessTransport(runtime.router)
client = IntentusClient(transport)

# Create MCP adapter
adapter = MCPAdapter(client)

# Handle MCP tool call
def handle_tool_call(tool_name: str, parameters: dict) -> dict:
    mcp_request = {
        "tool": tool_name,
        "parameters": parameters
    }
    return adapter.handle_mcp_request(mcp_request)

What the Adapter Does

1. Receives MCP tool call → Standard MCP format
2. Converts to IntentEnvelope → IntentusNet format
3. Routes through runtime → All guarantees apply
4. Records execution → Auditable, replayable
5. Returns MCP response → Standard MCP format

Benefits

For MCP Users

• Replay: Reproduce exact tool outputs for debugging
• Audit: Complete execution records
• Policy: Fine-grained access control
• Recovery: Handle crashes gracefully

For Tool Developers

• Same tools: No changes to existing MCP tools
• Added guarantees: Recording, policy, routing
• Incremental adoption: Add IntentusNet to existing MCP setup

Quick Example

from mcp import Server
from intentusnet import IntentusRuntime
from intentusnet.transport.mcp import MCPAdapter

# Set up IntentusNet
runtime = IntentusRuntime(enable_recording=True)
adapter = MCPAdapter(IntentusClient(InProcessTransport(runtime.router)))

# MCP server with IntentusNet backing
server = Server()

@server.tool("search_documents")
async def search_documents(query: str, limit: int = 10):
    # Route through IntentusNet
    result = adapter.handle_mcp_request({
        "tool": "search_documents",
        "parameters": {"query": query, "limit": limit}
    })

    # IntentusNet provides:
    # - Deterministic routing to search agent
    # - Execution recording
    # - Replay capability

    return result

Compatibility

MCP Feature	IntentusNet Support
Tool calls	Full
Tool discovery	Via agent registry
Context passing	Mapped to IntentContext
Streaming	Not currently supported
Cancellation	Not currently supported

Next Steps

• Integration Patterns — Detailed integration examples
• Protocol vs Runtime — Understanding the distinction