lex-foundry

Legion Extension for Azure AI Foundry (formerly Azure AI Studio). Provides runners for the Azure AI Foundry REST API — workspace connections, online endpoint deployments, and the model catalog.

Purpose

Wraps the Azure AI Foundry management and workspace APIs as named runners consumable by any LegionIO task chain. Use this extension when you need to manage Azure ML workspace resources (connections, deployments, models) within the LEX runner/actor lifecycle. For inference against deployed models, use lex-azure-ai instead.

Installation

Add to your Gemfile:

gem 'lex-foundry'

Authentication

All methods require an Azure AD / Entra ID Bearer token passed as token:.

Usage

Standalone Client

client = Legion::Extensions::Foundry::Client.new(
  token:           'your-bearer-token',
  endpoint:        'my-workspace',
  subscription_id: 'sub-id',
  resource_group:  'rg-name',
  workspace:       'ws-name'
)

client.list(subscription_id: 'sub-id', resource_group: 'rg', workspace: 'ws')

Runners Directly

Legion::Extensions::Foundry::Runners::Models.list(token: 'tok', endpoint: 'my-workspace')
Legion::Extensions::Foundry::Runners::Deployments.list(token: 'tok', endpoint: 'ep', subscription_id: 'sid', resource_group: 'rg', workspace: 'ws')
Legion::Extensions::Foundry::Runners::Connections.list(token: 'tok', endpoint: 'ep', subscription_id: 'sid', resource_group: 'rg', workspace: 'ws')

API Coverage

Runner	Methods
Connections	list, get, create, delete
Deployments	list, get, create, delete
Models	list, get

API Version

Default: 2024-10-01-preview. Pass api_version: to override.

Related

• lex-azure-ai — Azure OpenAI inference (chat completions, embeddings)
• extensions-ai/CLAUDE.md — Architecture patterns shared across all AI extensions

License

MIT