I started with a real task: let an AI coding agent containerize a Product Catalog app, and make it pick Docker Hardened Images (DHI) for its base layer by talking to the DHI MCP server at https://dhi.io/mcp. The first version was a pile of imperative sbx flags
What if an AI agent could run an A/B test all by itself? This tutorial walks you through building one on Docker Cloud Sandboxes.
More than 30K technologists from 169 countries have shared their thoughts with us, and now it’s time to share everything we’ve learned with you. So let’s dive into this year’s Annual Developer Survey and where Docker stood in the results
AI agents are no longer just chat assistants. A year ago, most of us used AI to answer questions or suggest a few lines of code. Today, agents write whole features, run shell commands, install packages, call APIs and push changes to real systems. Many of them do this with
Ask a developer what Docker is for and you still get "it packages my app." That answer was complete five years ago. Here is one morning with one app, and five Docker capabilities most people have not tried yet: Gordon, Testcontainers, Scout, Hardened Images, and Sandboxes.
AI coding agents now run for a long time. A single task can take hours: a refactor, a full test run, a migration. The question is no longer whether the model can finish the work, but where that work should run. Your laptop is fine for the first few minutes,
Giving an agent a filesystem, a shell, and a network hands generated code the same reach your laptop has. Here's how to package DeepAgents as a Docker Sandbox Kit that runs fully isolated against a local Docker Model Runner, with no cloud credentials and a one-command install.
Docker. Kubernetes. Agentic AI.
I’ve been in the Docker community since 2015. I’ve written more docker run tutorials than I can count, and I still get people in the Collabnix Slack asking me which base image to use for their Spring Boot app. That question is still valid. But it is no
An AI-BOM is a machine-readable inventory of the models, datasets, prompts and agents your AI system depends on. Here is how it extends the SBOM you already ship
Docker Sandboxes 0.39 lets Claude Code's /remote-control run inside a microVM. Drive an isolated agent from your phone while your host filesystem stays out of reach. Here is how to turn it on, plus the credential requirement that trips up most first attempts.
A hands-on tour of the MCP gateway, kit spec v2, structured startup progress, per-sandbox network deny rules, and daemon restart introduced in Docker Sandboxes (sbx) 0.38.0.
Claude Code, like most agentic coding tools, asks for permission before it does anything consequential: editing a file, running a shell command, installing a package, or making a network call. That permission system is a safety rail. 💡"Yolo mode" is the community nickname for turning that rail off
Rest isn't a reward. It's a part of how we work sustainably. Remote Life, Done Right ~ Docker
A technical deep dive into how Docker Sandboxes shares agent skills - the import flow, the persistent store, per-platform paths, the read-write trust boundary, and how it differs from shipping skills through a kit.
Here's a quick 9 tips for Sandboxing your AI agents with Docker Sandboxes Kits
Everything you want to know about the latest Docker sbx 0.35.0
Docker Model Runner supports three inference engines namely llama.cpp, vLLM, and Diffusers. Each engine has different strengths, supported platforms, and model format requirements. This blog article helps you in making a decision to choose the right engine and configure it for your use case.
Most developers running Claude Code today run it directly on their host machine. That means the agent shares an environment with your SSH keys, your AWS credentials, your browser sessions, and every other project on your laptop. A container helps, but a container is not a sandbox. A shared kernel
💡Updated for sbx v0.37.0-rc1. The SSH feature is still experimental and the command surface has moved since the first nightly builds. The examples below were re-run end to end on my own machine against v0.37.0-rc1, and every command here is verified working. If
