What problem does it solve? Running untrusted or multi-tenant code with strong isolation, long-lived sessions, and fast startup is hard with standard serverless functions. This Skill guides you through building, running, and operating AWS Lambda MicroVMs — Firecracker-isolated compute environments that boot from memory and disk snapshots, suspend and resume with state preserved, and live up to 8 hours. ## Core Features & Use Cases - Image build and snapshot workflow: Package a Dockerfile-based app in S3, create a MicroVM image with lifecycle hooks (/ready, /validate), and launch instances that resume from snapshots in seconds. - Full lifecycle operations: Run, suspend, resume, and terminate MicroVMs via CLI, mint scoped auth tokens, and configure idle policies with auto-resume. - Networking and security: Configure ingress/egress network connectors, VPC egress, shell access via WebSocket, IAM build/execution roles, and per-token port restrictions. - Use Case: Build an AI agent code-execution sandbox where each session gets a fresh isolated MicroVM, suspends between turns to save cost, and resumes with full state intact. ## Quick Start Ask the assistant to walk you through packaging a Flask app, creating a Lambda MicroVM image, running a MicroVM, and calling its endpoint with an auth token.