skills.rest
missBerg
missBergCOMMUNITY
Explore all skills by this author →
GitHub

aigw-contrib-architecture

Clarify Envoy AI Gateway architecture and two-level ExtProc flow.

3|Updated Mar 8, 2026
Search Tags:#architecture#translation#controller#crd#ai-gateway#envoy#extproc
One-click install
npx skills add https://github.com/missBerg/envoy-skills --skill aigw-contrib-architecture
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aigw-contrib-architecture
Source: https://github.com/missBerg/envoy-skills/tree/main/ai-gateway/contributors/skills/aigw-contrib-architecture
Command: npx skills add https://github.com/missBerg/envoy-skills --skill aigw-contrib-architecture

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Clarifies Envoy AI Gateway architecture and two-level ExtProc flow.

Core Features & Use Cases

  • Two-level ExtProc design (Router ExtProc before routing, Upstream ExtProc after routing) enables model-based routing and seamless schema translation.
  • File-based control-to-data-plane communication (controller writes YAML to a Secret, ExtProc watches the file) supports reproducible testing and local development.
  • CRDs and reconcilers map AI backends, routes, and policies to Envoy Gateway configurations, enabling end-to-end AI traffic management.

Quick Start

Review the architecture diagram and CRD layout to start tracing data and control flow.

Frequently Asked Questions about aigw-contrib-architecture

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How does the Envoy AI Gateway architecture handle routing and schema translation?▼

The Envoy AI Gateway architecture uses a two-level ExtProc design: a Router ExtProc before routing for model-based selection, and an Upstream ExtProc after routing for seamless schema translation.

How do controllers communicate with ExtProc sidecars in the AI Gateway data plane?▼

Controllers communicate with ExtProc sidecars using a file-based approach where the controller writes YAML configurations to a Secret, and the ExtProc watches this file to enable reproducible testing and local development.

How do CRDs map AI backends and routes to Envoy Gateway configurations?▼

Custom Resource Definitions and reconcilers map AI backends, routes, and policies directly to Envoy Gateway configurations, enabling end-to-end AI traffic management across the data plane.

What is the two-level ExtProc flow in Envoy AI Gateway deployments?▼

The two-level ExtProc flow consists of a Router ExtProc that executes before routing to enable model-based routing, and an Upstream ExtProc that executes after routing to handle schema translation.

Do I need to understand CRDs and reconcilers to navigate the envoyproxy/ai-gateway repository?▼

Yes, understanding the CRD layout, reconcilers, and key design types is required to effectively trace data and control flow within the envoyproxy/ai-gateway repository structure.

Can I use file-based control-to-data-plane communication for local AI Gateway development?▼

Yes, the AI Gateway supports file-based control-to-data-plane communication where controllers write YAML to a Secret and ExtProc watches the file, specifically enabling reproducible testing and local development.

Related Skills

Explore More Software Engineering Skills→
missBergmissBerg
3

eg-version

Identify Envoy Gateway versions and assess Gateway API upgrade readiness.

Community
Advanced
inernoroinernoro
2

flow-trace

Trace data and control flow to generate an end-to-end path diagram.

Community
Advanced
HyperscapeAIHyperscapeAI
91

api-gateway-configurator

Configure API gateways with rate limiting and JWT authentication.

Official
Advanced
doanchienthangdevdoanchienthangdev
4

API Gateway Patterns

Configure Kong and Traefik API gateway patterns for routing, security, and traffic management.

Community
Advanced