{"slug": "what-s-new-in-docker-sandbox-v0-53-0", "title": "What's New in Docker Sandbox v0.53.0?", "summary": "Docker released Sandboxes v0.53.0, a tool that runs AI coding agents inside an isolated microVM with its own Linux kernel, a private Docker daemon and a network proxy that enforces an allowlist, according to a hands-on review by Ajeet Raina. In testing, the sandbox's proxy returned 200 for github.com and 403 for an unlisted host under the default \"balanced\" policy, and the first launch pulled its kit image in about 26 seconds on an 18 CPU / 18 GiB allocation. Raina flagged the mounted workspace as the main residual risk, since an agent can modify files such as .git/hooks and CI config that do not appear in git diff.", "body_md": "# What's New in Docker Sandbox v0.53.0?\n\nGiving Your AI Agents a Safe Place to Play\n\nAI coding agents are great right up until they aren't. \"YOLO mode\" (letting the agent run commands without asking permission) is where the real productivity lives, but it's also where a single bad `rm -rf` or a sketchy `curl | sh` can ruin your afternoon (or your laptop).\n\nDocker Sandboxes (`sbx`) solves this the Docker way: it drops the agent into an isolated **microVM** with its own kernel, its own Docker daemon, and a network proxy that only lets through the traffic you allow.\n\nI spent some time putting the brand-new [**v0.53.0**](https://docs.docker.com/ai/sandboxes/release-notes/?ref=ajeetraina.com#0530) release through its paces. Here's what I found.\n\n## Installing\n\nOn macOS it's a one-liner, and notably, **Docker Desktop is not required**:\n\n```\nbrew install docker/tap/sbx\nbash\n$ sbx version\nsbx version v0.53.0 2c62838...\n```\n\nBefore running anything, `sbx diagnose` is a great sanity check. It walks the whole stack: CLI binary, daemon health, hardware virtualization, filesystem tooling, auth, and network policy:\n\n```\nINSTALLATION\n  ✓ CLI binary — found\n  ✓ Daemon — healthy (v0.53.0)\nPLATFORM\n  ✓ Virtualization — supported (kern.hv_support is 1)\nCONNECTION\n  ✓ Authentication — authenticated\n  ✓ Global network policy — initialized\n\n  17 passed · 3 skipped\n```\n\n## Spinning up a sandbox\n\nI created a throwaway project and launched a bare `shell` sandbox into it:\n\n```\nmkdir -p /tmp/sbx-053-test && cd /tmp/sbx-053-test\ngit init && echo 'print(\"hello from a microVM\")' > app.py\n\nsbx create shell --name blog-test /tmp/sbx-053-test\n```\n\nFirst launch pulls the kit image; mine was ready in about **26 seconds**, and subsequent runs are cached. The CLI tells you exactly what you're getting: the image digest, the mounted workspace, and the resources allocated (18 CPU / 18 GiB on my machine).\n\n## The four isolation layers, verified by hand\n\nThis is the part that matters, so I tested each boundary rather than taking the docs' word for it.\n\n### 1. A separate kernel\n\n``` bash\n$ sbx exec blog-test uname -a\nLinux blog-test 7.0.14 #1 SMP PREEMPT ... aarch64 GNU/Linux\n```\n\nThat's a full Linux guest kernel running on my macOS host via the hypervisor, not a shared namespace. The agent runs as an unprivileged `agent` user (uid 1000) but has `sudo` inside the box *by design*. The security boundary is the VM, not in-guest permissions.\n\n### 2. The workspace is mounted, and writes flow both ways\n\n``` bash\n$ sbx exec blog-test sh -c 'ls /tmp/sbx-053-test && cat app.py'\nREADME.md  app.py\nprint(\"hello from a microVM\")\n```\n\nWrite a file inside the sandbox and the host sees it instantly:\n\n``` bash\n$ sbx exec blog-test sh -c 'echo \"written inside the microVM\" > from-sandbox.txt'\n$ cat /tmp/sbx-053-test/from-sandbox.txt\nwritten inside the microVM\n```\n\n⚠️ **The workspace is the main residual risk.** The agent can touch anything in the mounted directory, including `.git/hooks/`, CI config, and build scripts, which don't show up in `git diff`. Worth a post-session glance.\n\n### 3. A private Docker daemon inside the sandbox\n\nThis one is my favorite. The sandbox ships its *own* Docker Engine, fully isolated from the host's:\n\n``` bash\n$ sbx exec blog-test docker version --format '{{.Server.Version}}'\n29.9.3\n\n$ sbx exec blog-test docker run --rm hello-world\nStatus: Downloaded newer image for hello-world:latest\nHello from Docker!\n```\n\nYour agent can build and run test containers all day long without ever touching your host daemon.\n\n### 4. A network proxy that actually enforces policy\n\nWith the default **balanced** policy, I hit an allowed host and a random unlisted one:\n\n``` bash\n$ curl -s -o /dev/null -w \"%{http_code}\\n\" https://github.com\n200\n$ curl -s -o /dev/null -w \"%{http_code}\\n\" https://www.random-unlisted-site-xyz.com\n403\n```\n\nAnd `sbx policy log` shows exactly why. Every decision is auditable:\n\n```\nBLOCKED REQUESTS\n  www.random-unlisted-site-xyz.com:443   No matching allow rule (default deny)\n\nALLOWED REQUESTS\n  registry-1.docker.io:443\n  github.com:443\n```\n\nNotice that `registry-1.docker.io` appears in the allow list. That's the `hello-world` pull from the test above, routed through the proxy. Nothing reaches the network without going through the policy engine.\n\n## What's new in v0.53.0\n\nA few highlights stood out beyond the core isolation story.\n\n**`sbx df`: local disk accounting.** A new command that finally answers \"where did all my disk go?\" with a `docker system df`-style breakdown and a reclaimable estimate:\n\n``` bash\n$ sbx df\nTYPE            TOTAL   ACTIVE   SIZE      RECLAIMABLE\nImages          113     19       50.29GB   17.43GB (35%)\nContainers      46      1        12.93GB   12.93GB (100%)\nLocal Volumes   162     162      4.733GB   0B (0%)\n```\n\n**Claude in Chrome access.** Enable the `claude.chrome` setting and sandboxed Claude Code can drive Claude in Chrome using your subscription OAuth login (start it with `--chrome`). API-key auth alone won't cut it. This is subscription-gated.\n\n**Automatic loopback port publication for kits.** Kits granted `ports.publish` with `approval: none` can grab an auto-assigned loopback port without a prompt (fixed ports still require approval).\n\n**Resilience after network changes.** Stalled outbound connections now close after ~30s so clients can reconnect cleanly. Handy when you connect a VPN mid-session. There's also a specific fix for TLS-inspecting proxies (Zscaler, Palo Alto, F5) that were rejecting ML-DSA certificate handshakes.\n\nA couple of **breaking changes** worth noting if you're upgrading:\n\n- The built-in `gemini` agent name is**retired** . Existing Gemini sandboxes keep running but can't be`sbx move` 'd.\n- `sbx run <agent> -- <args>` now preserves the image entrypoint args to match cloud behavior, so don't double up on subcommands.\n- `DOCKER_SANDBOXES_USE_DHI` is gone; use`sbx settings set platform.images.useDHI true` .\n\nNote: v0.52.0 was skipped, so these notes cover everything since v0.51.0.\n\n## Cleaning up\n\n```\nsbx rm -f blog-test\n```\n\nDone. State gone, nothing left behind on the host.\n\n## Verdict\n\nv0.53.0 feels mature. The isolation story is exactly what it claims to be (I verified all four layers in a couple of minutes), and the new `sbx df` command plugs a real ergonomic gap. If you're letting coding agents run unattended, this is the difference between \"autonomous\" and \"reckless.\" It runs great without Docker Desktop too, which makes it an easy add to any workflow.\n\nGive it a spin:\n\n```\nbrew install docker/tap/sbx\nsbx login\nsbx run claude ~/my-project\n```\n\n📚 Docs: [https://docs.docker.com/ai/sandboxes/](https://docs.docker.com/ai/sandboxes/?ref=ajeetraina.com)\n\n📦 Releases: [https://github.com/docker/sbx-releases/releases](https://github.com/docker/sbx-releases/releases?ref=ajeetraina.com)", "url": "https://wpnews.pro/news/what-s-new-in-docker-sandbox-v0-53-0", "canonical_source": "https://www.ajeetraina.com/whats-new-in-docker-sandbox-v0-53-0/", "published_at": "2026-11-11 17:25:46+00:00", "updated_at": "2026-11-22 15:49:28.228677+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-tools", "developer-tools"], "entities": ["Docker", "Docker Sandboxes", "sbx", "Ajeet Raina", "Docker Desktop", "GitHub"], "also_reported_by": [], "takeaway": "Docker Sandboxes v0.53.0 isolates AI coding agents in a microVM with a separate kernel, a private Docker daemon and a default \"balanced\" network proxy that returned 403 for unlisted hosts, but the mounted workspace stays writable, so review files like .git/hooks and CI config after each session.", "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/what-s-new-in-docker-sandbox-v0-53-0", "markdown": "https://wpnews.pro/news/what-s-new-in-docker-sandbox-v0-53-0.md", "text": "https://wpnews.pro/news/what-s-new-in-docker-sandbox-v0-53-0.txt", "jsonld": "https://wpnews.pro/news/what-s-new-in-docker-sandbox-v0-53-0.jsonld"}}