Running Sandboxed Coding Agents in your CI/CD Pipeline

In this hands-on workshop, you'll build sandboxed execution environments for AI agents using AWS Lambda MicroVMs — Lambda's newest compute primitive that combines Firecracker VM-level isolation with serverless scaling and zero infrastructure management.

You'll start in Module 1 with the fundamentals every other module builds on: create a custom MicroVM image with a Dockerfile, launch your first MicroVM, and interact with it over HTTP. From there, the use-case modules are independent - pick the ones that interest you and skip the rest:

• Module 2 - Event-Driven Coding Agents: a pull request triggers an ephemeral MicroVM that runs an AI coding agent over the diff and posts a review back.
• Module 3 - Ephemeral CI/CD Runners: a per-job runner launches on a push, runs the pipeline in a clean MicroVM, reports the result, and disappears.
• Module 4 - Multi-tenant SaaS: each customer gets their own MicroVM behind a control plane that launches, caches, and routes to per-tenant instances.

Along the way you'll work with the core building blocks: MicroVM images, snapshots for fast startup, network connectors, and the launch/suspend/resume/terminate lifecycle.

Prerequisites: Basic AWS familiarity, serverless concepts, CLI experience. AWS account provisioned by the workshop.

Duration: ~2.5–3 hours (3 modules, 30–45 minutes each).