ACP: The AI Agent Escape Hatch

Posted on Tue 12 May 2026 in AI Engineering • Tagged with ai-agents, acp, sandboxing, multi-agent, developer-tools, nemoclaw, openclaw


The Walled Garden Problem

Modern AI agents are powerful — but deliberately caged.

Tools like NemoClaw run AI subagents inside sandboxes: isolated environments where internet access is blocked, your local files are untouchable, and nothing persists between sessions. This is great for security. It's terrible when you actually need your AI …


Continue reading

Build NemoClaw and Connect to Your Sandbox

Posted on Tue 12 May 2026 in Developer Tools • Tagged with nemoclaw, cli, sandbox, ai-agents, build-from-source, openclaw

Build NemoClaw & Connect to Your Sandbox

So you want to run your own AI agent sandbox. Not the hosted, hand-holdy version — the real thing, built from source, running on your machine.

This is the guide I wish I had.


What Is NemoClaw?

NemoClaw is a sandbox manager for AI agents …


Continue reading

How NemoClaw Works Under the Hood

Posted on Tue 12 May 2026 in AI Engineering • Tagged with nemoclaw, acp, security-model, ai-agents, config, subagents, openclaw, sandbox

How NemoClaw Works Under the Hood

You've built the sandbox. You've sent a Slack message from inside it. Now let's talk about why things work the way they do — and why certain things that seem like they should work, don't.

This post is about the architecture. Once you understand it …


Continue reading