Docker, Inc

Docker Captains are the community members who build, teach, and share what they know. Captain Steve Buchanan has now turned that into a course. Docker in Practice: Build, Debug, and Deploy Containers is the first course in the new Docker Foundations Professional Certificate on LinkedIn Learning, and it also anchors the DevOps learning paths.  Start the path, pass the final exam, and earn your certificate: https://lnkd.in/g_yADX_C

The isolation question for AI agents is getting more specific. Not 'is it sandboxed?' but 'what exactly does the boundary hold, and what can get past it?' Sandbox Review on how to answer that: what the isolation spectrum actually looks like, what Docker's own Cloud Sandboxes launch says about the architecture, and what major coding agents are doing about isolation in production today: https://lnkd.in/eZXN-fvU

An agent that breaks something isn't broken. It's doing exactly what it was built to do, just not what you intended. Docker’s Eric Jia and Srini Sekaran break down the five main agent security risks: prompt injection, tool misuse and poisoning, excessive autonomy, credential exposure, and data exfiltration. Every one of them has produced documented real-world incidents, and all of them trace back to the same thing: the agent can act, so containment and oversight matter more than the model. Read more →

Our engineers use Docker Sandboxes for their own projects too. This one's a Claude Code plugin for watching YouTube while agents work in the background, with a Docker Sandbox Kit. New sign ups get $250 in Cloud Sandboxes credit through October 31 to try it yourself: https://lnkd.in/eRRqjgRU

Thank you, WeAreDevelopers! Spending time with the developer community was the highlight for us. The conversations, the ideas, and the energy in the room are what make this event so special.

The 2026 Stack Overflow Developer Survey is out. We’re honored that Docker continues to be the most used and most desired tool in cloud development. The results also make it clear that trust is the biggest gap in the agentic era. Only 7% of developers trust AI output for important work decisions, and nearly half (48%) trust it only when they can verify it themselves. In a new blog post, Docker’s CMO Terri Avnaim explains how we’re building on a decade of developer trust and extending it from containers to sandboxes. Read now:

Here’s what our COO is doing with sandboxes. What could you do?  Try Cloud Sandboxes this month and get a $250 credit: https://lnkd.in/eRRqjgRU

Agents that can edit files and run commands can also game their own evals. You need an eval framework that runs each task, lets the agent work, checks the results, and collects the scores and logs. You also need an environment for every task that’s clean and isolated, so evaluating agents is safe and consistent. With Harbor and Docker Sandboxes, you get both. Harbor now runs every eval task in Docker Sandboxes. This means isolated microVMs, network policies for fair and reproducible runs, and the same setup locally or at scale in the cloud. Learn more here:

Too much Python boilerplate to get an agent running? Better Stack covered how to use Docker Agent to simplify it: one YAML file, one command, agents talking to each other with nothing more than a URL. The video walks through what Docker Agent is, demos two agents collaborating on a real debugging scenario, and ends with GPT-6 Luna and Claude Haiku playing battleship against each other. No framework code anywhere. Watch here → https://lnkd.in/eJ5ynzAq