Codegen agents are becoming standard for both local and remote workloads. As their capabilities grow, so does the blast radius when their experiments go wrong: deleted files, leaked secrets, corrupted state, and in adversarial cases, host compromise.Sandboxes are the emerging answer. Think of them like a laptop for agents: close the lid and the environment suspends, open it and it resumes instantly with full state preserved. Hardware-level isolation with sub-second startup, giving agents somewhere to run arbitrary code without touching the host.With a sandbox in place, the next question is what to put in front of it. A single agent with a big toolbox and one sandbox is the obvious starting point…and it works, until it doesn’t. Context windows balloon, the model second-guesses its own tool choices, and one prompt ends up trying to do reviewer, refactorer, and test-writer all at once.Specialized agents are a better unit of composition: each one narrow in scope, easier to evaluate, easier to swap out, and easier to reason about when something goes wrong. But the hard part isn’t building one agent; it’s getting two of them to share enough context to collaborate without stepping on each other.This full-day, hands-on workshop walks attendees through the complete journey: from a single-prompt prototype to a sandboxed, multi-agent system with shared context. We will build it using the OpenAI Agents SDK (Apache 2.0, LLM-agnostic) wired to a cloud sandbox via MCP. The pattern is vendor-agnostic and works with any LLM and sandbox provider supported by the SDK. Teams with data sovereignty requirements can swap in self-hosted open-weight models such as Llama 3 or Mistral served via vLLM or Ollama.By the end of the day, you’ll have:– A working agent built on the OpenAI Agents SDK, running locally – That same agent executing code inside an isolated Firecracker microVM sandbox, wired up via MCP – A working use case: an automated code review agent that lints, analyzes, and gives feedback on real pull requests – Another agent collaborating with the first through shared context, demonstrating a multi-agent pattern – A clear mental model for when to reach for microVM sandboxes versus containers versus serverless, and what production-readiness looks like from here## Audience Developers building AI agents, platform and infrastructure engineers supporting them, and technical leads deciding what their agentic stack should look like.## Prerequisites Comfort with Python or TypeScript and basic command-line work is expected. Laptops, GitHub account, LLM provider API key, sandbox provider API key required. A setup guide will be sent in advance.## OutlineHour 1: Risks of agent-generated code. Containers vs VMs vs serverless. Introduction to tools used in workshop. Setup Hour 2: Build your first agent. Run it locally. Hour 3: Add a sandbox. Build an image. Connect the agent to the sandbox with it using MCP. Move code execution off the laptop into a cloud sandbox. Hour 4: Build a concrete use case: code review agent. Hour 5: Go multi-agent. Introduce a second agent, share context between agents. Hour 6: Ship time: deploy, add tracing, debug. Discuss production issues. Q&A.## DisclosureThis workshop uses sandboxes from blaxel.ai, where the presenter works as an independent consultant. Free signup and credits are available for new users. Other sandbox providers are also supported, users can choose a different one if they wish but due to time constraints the presenter may not be able to support all combinations. All patterns and concepts in this workshop generalize to other sandbox providers and other LLMs.