Skip to main content

Docker and CI/CD

Beginner explanation​

Docker packages the app and its runtime. CI/CD automates test, build, and deployment steps so releases are repeatable.

Production explanation​

Agentic apps often depend on more than a Node server: vector databases, background jobs, browser runtimes, and provider credentials. CI/CD should validate these dependencies before production rollout.

Real-world enterprise example​

A QA agent API runs in containers, its browser automation dependency is validated in CI, and staging deployment runs smoke tests before production promotion.

Mermaid diagram​

TypeScript example​

export function healthcheck() {
  return {
    status: 'ok',
    providerConfigured: Boolean(process.env.OPENAI_API_KEY),
  };
}

Python example​

def smoke_test() -> bool:
    return True

Common mistakes​

  • assuming local success means container success
  • deploying prompt or retrieval changes without a staging gate
  • no smoke tests for external dependencies
  • not documenting required environment variables

Mini exercise​

Write the CI stages for one project and identify which checks should block production.

Project assignment​

Document a container plan and CI gate list for your chosen project.

Interview questions​

  • What should a smoke test validate for an AI application?
  • Why do prompt changes deserve release discipline?
  • What dependencies are easiest to miss in CI?

Monetization angle​

Productionization work is often easier to sell than greenfield experimentation because it directly reduces risk and improves delivery confidence.