Skip to main content
Learn about the automated CI/CD workflows that streamline the development, testing, and deployment of agents on the Agentic AI platform.

​
What You’ll Learn

Build and Publish

Automated build, test, and publish processes

Pull Request Check

PR validation, quality checks, and security scans

Release

Production release and publishing

Release Draft

Create release drafts with quality checks

Hotfix Release

Automatic hotfix release drafts

Repository Provisioning

Initial repository setup and configuration

Reusable Components

Common workflows and actions reference

Configuration

Configuration files, secrets, and templates

​
CI/CD Pipeline Overview

The Agentic AI platform uses a comprehensive CI/CD pipeline that automates:
  1. Repository Provisioning: Initial setup when cloning from template
  2. Pull Request Validation: Quality checks, security scans, and testing on PRs
  3. Build and Publish: Automated build and artifact publishing on merge
  4. Release Management: Create release drafts and publish production releases
  5. Hotfix Handling: Emergency release process for critical fixes
The workflows leverage reusable GitHub Actions workflows from backbase-common/gc-ai-workflows for consistency across all agent projects.

​
Reusable Workflows and Actions

All workflows use standardized reusable components from backbase-common/gc-ai-workflows:

​
Reusable Workflows

  • build-publish.yaml: Complete build, test, and publish pipeline
  • pull-request-check.yaml: Comprehensive PR validation workflow

​
Reusable Actions

  • setup-project: Project environment setup with Python and uv
  • code-quality: Code quality checks (pylint, pytest, hadolint)
  • sonar-check: SonarCloud code analysis
  • security-check: Trivy vulnerability scanning
  • promptfoo-evaluation: LLM prompt evaluation
  • promptfoo-redteaming: Security and adversarial testing
  • build-docker: Docker image building
  • push-docker: Docker image publishing to Azure ACR
  • create-release-draft: Release draft creation with versioning
  • provision-python-project: Python project template provisioning
  • validate-pull-request-body: PR body validation
  • check-action-pinning: GitHub Actions security validation
  • resolve-metadata: Version and metadata resolution
  • setup-promptfoo: Promptfoo environment setup
See the Reusable Components page for detailed documentation.

​
Workflow Files Location

All CI/CD workflows are defined in .github/workflows/ directory in your agent repository:
  • build-publish.yaml - Build and publish artifacts on PR merge
  • pull-request-check.yaml - PR validation and quality checks
  • release.yaml - Production release workflow
  • release-draft.yaml - Create release drafts manually
  • hotfix-release-draft.yaml - Create hotfix release drafts
  • repository-provisioning.yaml - Initial repository setup
These workflows use reusable workflows from backbase-common/gc-ai-workflows and follow standardized CI/CD practices.

​
Workflow Stages

​
Repository Provisioning

Automated setup when creating a new agent from template, including project configuration and baseline versioning.

​
Pull Request Validation

Comprehensive validation including:
  • Code quality checks (pylint, pytest, hadolint)
  • Security scanning (Trivy)
  • SonarCloud analysis
  • Testing (Promptfoo evaluation, Redteam)
  • PR body validation

​
Build and Publish

Automated build and artifact publishing triggered on PR merge, with optional quality gates for main/develop branches:
  • Docker image building
  • Security scanning
  • Publishing to Azure Container Registry

​
Release Management

  • Manual release draft creation with quality checks
  • Automatic hotfix release drafts
  • Production release publishing with full quality gates

​
Benefits

  • Automation: Reduce manual errors and speed up delivery
  • Quality: Automated testing and validation at every stage
  • Consistency: Standardized processes across all agent projects via reusable workflows
  • Visibility: Clear feedback on build and deployment status
  • Security: Integrated security scanning and quality gates

​
Next Steps