Skip to content

Meaning Protocol Layer

Semantic governance for AI agents in regulated environments.

MPL is a lightweight protocol overlay that brings typed contracts, quality SLOs, and audit trails to AI agent communications—without replacing MCP or A2A.

  • Typed Contracts

    Semantic Types (STypes) provide versioned, schema-backed contracts for every message between AI agents.

    Learn about STypes

  • Quality SLOs

    Quality of Meaning (QoM) metrics measure and enforce semantic quality with configurable profiles.

    Explore QoM

  • Audit Trails

    BLAKE3 semantic hashing and provenance metadata provide tamper-evident audit trails for every interaction.

    See Audit Trails

  • Zero-Code Integration

    Deploy as a sidecar proxy alongside existing MCP/A2A infrastructure. No code changes required.

    Integration Modes

Start Here

Choose your path based on your role:

Role Start Here You'll Learn
Executive / CTO Why MPL How MPL unblocks AI deployment in regulated environments
CISO / Compliance Security Overview Compliance mapping, threat model, audit capabilities
Architect Architecture Protocol design, integration patterns, QoM system
Engineer Quick Start Install, configure, and validate in 5 minutes

Quick Start

# Install the CLI
cargo install mplx

# Start proxy pointing to your MCP server
mpl proxy http://your-mcp-server:8080

# Dashboard: http://localhost:9080
# Metrics:   http://localhost:9100/metrics

What you get immediately:

  • Traffic visibility for all MCP/A2A requests
  • Real-time metrics and dashboard
  • Schema learning from observed traffic

Full Quick Start Guide

Core Concepts at a Glance

graph LR
    A[Agent] -->|typed call| B[MPL Proxy]
    B -->|validate| C[Schema Registry]
    B -->|evaluate| D[QoM Engine]
    B -->|enforce| E[Policy Engine]
    B -->|forward| F[MCP/A2A Server]
    F -->|response| B
    B -->|envelope + report| A
Concept Description
STypes Versioned semantic type identifiers (e.g., org.calendar.Event.v1) backed by JSON Schema
QoM Six quality metrics: Schema Fidelity, Instruction Compliance, Groundedness, Determinism, Ontology Adherence, Tool Outcome Correctness
Envelope Message wrapper carrying payload, SType, semantic hash, provenance, and QoM report
AI-ALPN Capability negotiation handshake before work begins
Policy Engine Rule-based enforcement of organizational policies
Registry Versioned store of SType schemas, profiles, and assertions

SDKs

pip install mpl-sdk

from mpl_sdk import Client, Mode

async with Client("http://localhost:9443", mode=Mode.PRODUCTION) as client:
    result = await client.call("calendar.create", {"title": "Meeting"})
    assert result.valid
    assert result.qom_passed

npm install @mpl/sdk

import { MplClient, QomProfile } from '@mpl/sdk';

const client = new MplClient('http://localhost:9443');
await client.negotiate({
  stypes: ['org.calendar.Event.v1'],
  profile: QomProfile.Strict
});

const result = await client.validate({
  stype: 'org.calendar.Event.v1',
  payload: { title: 'Meeting', start: '2025-01-15T10:00:00Z' }
});

Python SDK Reference | TypeScript SDK Reference

Compliance at a Glance

Regulation MPL Control
SOX Semantic hashes + provenance for tamper-evident audit trails
GDPR Consent references in envelopes; policy engine for data handling
HIPAA SType patterns restrict PHI access; QoM thresholds enforce accuracy
EU AI Act QoM metrics for transparency; provenance for explainability

Full Compliance Mapping