Azure OpenAI Operations Guide

Validate Azure OpenAI integration behavior against the current plugin and model discovery surfaces shipped in this repository.

Environment

# Required for real runs
export AZURE_OPENAI_ENDPOINT="https://<your-resource>.openai.azure.com"
export AZURE_OPENAI_API_KEY="<key>"
# Optional: API version override (default: 2024-02-15-preview)
# export AZURE_OPENAI_API_VERSION=2024-10-21

# Local mock mode: export TRAIGENT_MOCK_LLM=true, or call
# traigent.testing.enable_mock_mode_for_quickstart() in tutorial code.
# Provider-specific *_MOCK env vars are ignored.

# Note: use `--provider azure` in the experiment scripts below.

Dependencies

pip install -e ".[integrations]"  # or: pip install -r requirements/requirements-integrations.txt

Validation Checks

pytest -q tests/unit/integrations/test_cloud_plugins.py -o addopts='' -k azure
pytest -q tests/unit/integrations/test_model_discovery.py -o addopts='' -k AzureOpenAIDiscovery

These checks verify that Azure-specific parameter overrides and model-discovery helpers still match the current code. This repository does not currently ship a standalone Azure example script.

Note: In Azure OpenAI, model is treated as the deployment name.