Skip to content

Deployment

The MPL proxy is a single Rust binary that sits between your AI agents and MCP/A2A servers. It can be deployed as a native binary, a Docker container, or a Kubernetes sidecar -- choose the method that matches your environment and operational maturity.

Deployment Options

MPL supports three primary deployment patterns, each suited to a different stage of your workflow:

graph LR
    A[Development<br/>cargo run / Docker] --> B[Staging<br/>Docker Compose]
    B --> C[Production<br/>Kubernetes + Helm]

Development

Run the proxy locally for rapid iteration:

cargo run --release --package mpl-proxy -- --config mpl-config.yaml

docker run -p 9443:9443 -p 9100:9100 \
  -v ./registry:/app/registry:ro \
  -v ./mpl-config.yaml:/app/mpl-config.yaml:ro \
  mpl-proxy:latest

Staging

Use Docker Compose to run the full stack locally -- proxy, demo server, and optional monitoring:

docker compose up -d

Production

Deploy with the Helm chart for autoscaling, health probes, network policies, and observability:

helm install mpl-proxy ./helm/mpl-proxy \
  --set mpl.mode=strict \
  --set mpl.requiredProfile=qom-strict-argcheck \
  --set transport.upstream=http://mcp-server:8080

Decision Matrix

Environment Method Complexity Features Guide
Development cargo run or Docker Low Hot reload, debug logging Docker
Staging Docker Compose Medium Full stack, metrics, dashboards Docker
Production Kubernetes + Helm High HPA, TLS, network policies, PDB, ServiceMonitor Kubernetes & Helm

Ports

All deployment methods expose the same set of ports:

Port Purpose Protocol
9443 Proxy (MCP/A2A traffic) HTTP
9100 Prometheus metrics HTTP
9080 Dashboard UI HTTP

Configuration

Regardless of deployment method, the proxy reads its configuration from mpl-config.yaml. Key sections:

Section Controls
transport Listen address, upstream target, protocol
mpl Mode, registry path, QoM profile, schema enforcement
observability Metrics port, log level, log format
routing (Optional) SType-based traffic routing

See the Configuration Reference for full details.

Deployment Guides

  • Docker Deployment

    Build, run, and compose the MPL proxy as a container.

    Docker Guide

  • Kubernetes & Helm

    Deploy with the Helm chart for production-grade infrastructure.

    Kubernetes Guide

  • Production Checklist

    Security, reliability, and observability requirements before going live.

    Production Checklist

Architecture Recap

In all deployment modes, MPL operates as a transparent reverse proxy:

graph LR
    Agent[AI Agent] -->|request| MPL[MPL Proxy :9443]
    MPL -->|validated| Server[MCP/A2A Server]
    MPL -->|metrics| Prom[Prometheus :9100]
    MPL -->|dashboard| Dash[Dashboard :9080]

The proxy intercepts traffic, validates it against the SType registry, evaluates QoM metrics, enforces policies, and forwards valid requests upstream. Invalid requests are either logged (transparent mode) or rejected (strict mode).