Docker, Inc

You’ve probably already hit some version of this: A hardened image breaks your build. An agent needs more access than you’re comfortable giving it. A local AI setup works until you try to move it toward production. A trusted package suddenly becomes part of a supply chain incident. These are not separate problems. They are part of the same shift happening across AI workflows, infrastructure, and software supply chains. The latest Docker Navigator issue connects the patterns teams are working through now, from sandbox isolation and local model workflows to hardened images and supply chain attacks. Read in full → https://bit.ly/4tSi3cm

If you’re running agents today, you probably already feel this tension: You want to give them freedom to operate, but you don’t want them accessing everything. Docker AI Governance is a new way to define that boundary. It combines sandbox controls, network policies, and MCP governance into a single system, so agents can execute real workflows while staying inside clearly defined limits. No per-device setup. No fragmented controls. Full visibility. Available now. Read → https://bit.ly/3PGfn3o

Ok, now, #Reachy Mini has a brain: it's Docker, Inc Agent in #SBX + #Qwen 3.5 4b (served by Docker Model Runner) And the week-end is not finished 🤓 https://lnkd.in/eksePcWb

Most GenAI projects do not break because the model is wrong. They break when you try to run them in the real world. The new book by Docker Captain Harsh Manvar, and Ajeet Singh Raina focuses on that gap. It explores what it actually takes to go from a working demo to something that can be containerized, scaled, and trusted in production. It walks through running LLMs, wiring up agents, and making those systems work with the platforms and tooling teams already rely on, like Kubernetes and MCP. Great read if you’ve been on the “it works on my machine” → environment drift → integration headaches pipeline. https://amzn.to/3PeXZmr

If you're in London for LeadDev or Tessl's AI DevCon, make sure you’ve got this on your calendar. Docker and Black Duck are hosting a free evening meetup on June 2 for developers thinking seriously about AI in production - specifically the security questions that don't get enough stage time at the big conferences. Limited spots (50 total). Real conversations. Drinks and food on us. Tickets here → https://luma.com/o9i3yqns

Custom Catalogs and Profiles are now generally available in Docker’s MCP Catalog and Toolkit! This is a major step toward making MCP more flexible, portable, and easier to manage. - Custom MCP Catalogs let organizations package and distribute trusted MCP servers, including internal and custom ones, as portable OCI artifacts. - MCP Profiles let developers switch between focused toolsets for coding, planning, research, and more, while persisting configuration and sharing workflows across teams. We’ve provided a guide to get you started, with steps on how to build and publish a trusted custom catalog from Docker’s MCP Catalog and your own servers →

Excited to back NanoCo (creators of NanoClaw) - congratulations to the whole team! 🎉

Deleted files. Leaked secrets. Broken configs. Agents taking actions nobody intended, at machine speed. How many times have you watched things start going off the rails and realized you’re now racing the agent instead of working with it? In the new 'AI Coding Agent Horror Stories' series, Ajeet Singh Raina breaks down what these failures actually look like in practice. This first issue focuses on isolation, containment, and the execution-layer controls teams are starting to rely on once agents begin acting autonomously. Read issue 1 →

Gordon is now generally available. Docker’s AI agent reads your running containers, logs, and Compose files. It already knows your environment. Gordon traces failures, proposes fixes, and takes action with your approval, at every step.  Built into Docker Desktop and the CLI. Free with every Docker account. Your coding assistant helps you write code. Gordon helps you ship it. Try it →