Blog — page 17 of 38
How to give the Claude Agent SDK a code execution sandbox
The Agent SDK ships with Bash, Read, Write, and Edit. All four run on the host process. That's perfect on your laptop and wrong the moment a user is on the other end.
The Best Sandboxes for Claude Agents in 2026
The Claude Agent SDK gives you Bash, Read, Write, and Edit — all running in your process. Deciding where they should run instead is the real architectural choice.
The Best NestJS Hosting Platforms in 2026
Half of NestJS's best features assume a process that stays alive. Deploy it to a serverless platform and you find out which half the hard way.
The Best Vue Hosting Platforms in 2026
The single most useful question about hosting Vue: does your build produce a dist folder, or a server? Everything else follows from the answer.
The Best AWS Fargate Alternatives in 2026
Most people looking for a Fargate alternative aren't unhappy with Fargate. They're unhappy with the task definition, the load balancer, and the bill for idle tasks.
The Best Heroku Postgres Alternatives in 2026
Heroku Postgres was the first managed Postgres most of us used, and the price per gigabyte is why most of us eventually left. Here's the field in 2026.
How to give a LangChain agent a code execution tool
PythonREPLTool runs the model's code in your process, with your environment variables and your filesystem. Here's the version you can actually put in front of users, in about thirty lines.
How to add code execution to the OpenAI Agents SDK
The Agents SDK makes the tool wiring trivial. What it doesn't decide for you is where the code runs, how long state survives, and who cleans up — which is where the interesting choices are.
How to give a CrewAI agent a code execution tool
CrewAI's built-in code interpreter wants Docker on the machine running your crew. Here's the version that doesn't, and what changes when several agents share one kernel.
How to add a code interpreter to the Vercel AI SDK
The AI SDK's tool() helper makes the wiring trivial. The interesting part is everything around it: where the code runs, how state survives a serverless request, and who kills the sandbox.
How to add a custom domain to your app
Two DNS records and a wait. The mechanics are simple; almost every problem is a DNS problem, and it's usually one of four specific ones.
How to debug a failed deployment
Deployment failures fall into about six categories, and the log that explains yours is almost never the one you opened first. Here's the order to read them in.
How to control sandbox lifetime: TTL, idle, and cleanup
The bill for agent infrastructure is rarely the work. It's the sandboxes nobody killed because the process that created them crashed on step four.
The best Northflank alternatives in 2026
Northflank is one of the more complete PaaS products out there, which makes 'what's the alternative' a harder question than usual. It depends entirely on which part you're actually replacing.
The best CockroachDB alternatives in 2026
CockroachDB solves a real and difficult problem. The question worth asking before you shop for alternatives is whether it's the problem you actually have.
The best platforms for running Playwright tests in 2026
A Playwright suite that takes four minutes locally and twenty-five in CI is not a Playwright problem. It's a place-to-run-them problem, and there are four distinct answers.
Fork-PR CI Without Getting Pwned: One microVM Per Job
A stranger opens a pull request and your CI happily runs their postinstall script next to your npm token. Here's the pwn-request pattern, why self-hosted runners make it worse, and the per-job microVM shape that ends it.
Per-Tenant Static Site Builds in microVMs
Your build fleet is a public code-execution API with extra YAML. A per-build microVM turns a hostile postinstall from a fleet-wide incident into one failed build.
Runnable Code Playgrounds in Your Docs, Safely
A Run button in your docs is a free, unauthenticated compute API wearing a documentation costume. Here's the threat model, the latency budget, and the per-run microVM architecture that survives launch week.
Firecracker on arm64 vs x86_64: What Actually Differs
Same API, same virtio-over-MMIO, same jailer. Different boot protocol, different interrupt controller, different clock — and snapshots that never cross the architecture line. Here's the honest diff.
Confidential computing for microVMs: SEV-SNP, TDX, and arm CCA explained
Ordinary virtualization protects the host from the guest. Confidential computing tries to protect the guest from the host — including from the operator holding the keys to the datacenter. Here's what the hardware actually does, what it very much does not do, and why you probably don't need it.
Linux capabilities, explained for sandboxing: five sets, one bounding ceiling, and the bits that are still root
Capabilities split root into about forty bits so you can hand out one instead of all of them. Worth doing, cheap, real — and about a dozen of those bits are still root wearing a lanyard.
io_uring and sandbox security: the fast path that walks around your syscall filter
We invented an interface that lets you do I/O without making syscalls, and then acted surprised when the syscall filter stopped seeing the I/O. io_uring is a genuine performance win and a genuinely awkward sandbox problem — here's the mechanism, the mitigations, and why a guest kernel changes the math.
Best Jupyter notebook hosting platforms in 2026
JupyterHub, Colab, Deepnote, Hex, Databricks, SageMaker Studio, Modal, Binder, Codespaces, and PandaStack — judged on who runs the kernel, what it costs while idle, and whether you can safely run somebody else's notebook.