san360/agent-devops

This repository provides a complete CI/CD lifecycle for an Azure AI Foundry prompt agent. It's aimed at engineers and teams (ML engineers, DevOps, platform owners) building and operating prompt-driven agents who need reproducible versioning, evaluation gates, and rollback. Key features: - Versioned agent configs and system prompts, golden eval dataset, evaluator config, and unit tests. - Scripts to bootstrap Azure/GitHub infra (resource groups, Foundry projects, App Registration, federated credentials), deploy agents to TEST/PROD, run evaluations, compare models, and teardown. - Lifecycle demo with three phases: Phase 1 (web_search), Phase 2 (add code_interpreter + data analysis), Phase 3 (model upgrade to gpt-4.1) — each implemented as PRs that trigger CI. - CI workflows: PR evaluation gate (deploy to TEST, run Foundry evals, post results), deploy-to-PROD on merge, and daily monitoring for drift. - IaC (Bicep), artifacts for rollback, and automation for model availability validation and GitHub variable management.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.