MAS-Aviary

Multi-Agent System for Aviary Aircraft Design Optimization

MAS-Aviary is a multi-agent framework that orchestrates LLM-powered agents to perform aircraft design optimization using NASA’s OpenMDAO Aviary toolkit. It supports eight distinct coordination strategies formed by combining three organizational structures with three operational handlers.

The Aviary MCP server backend is available at github.com/Jezemba/Aviary.

Contents

• Quickstart
  • Prerequisites
  • Installation
  • Starting the Aviary MCP Server
  • Running a Single Optimization
  • Understanding the Output
  • Running the Dashboard
  • Next Steps
• Architecture
  • Overview
  • Organizational Structures (Strategies)
  • Operational Handlers
  • Strategy x Handler Combinations
  • Blackboard Communication (Networked)
  • Orchestrator Delegation Flow
  • Sequential Pipeline Stages
  • Agent Lifecycle and MCP Interaction
• Configuration
  • aviary_run.yaml
  • aviary_sequential.yaml
  • aviary_networked.yaml
  • aviary_graph.yaml
  • aviary_staged_pipeline.yaml
  • eval_thresholds.yaml
• API Reference
  • Coordination
  • Strategies
  • Handlers
  • Tools
  • LLM
  • Logging

Indices and Tables

• Index

• Module Index

• Search Page