Topic
A security technique that runs untrusted or AI-generated code in isolated environments, limiting its ability to affect host systems or data.
OpenAI's Agents API, in public beta since DevDay 2026, runs sessions, orchestration, context compaction and recovery for developers' agents. Teams that wrote their own loop can hand that state to OpenAI and keep their tools and their choice of sandbox.
Reality
A dev.to walkthrough of Microsoft Foundry's Agent Service pulls the tool declaration apart from the lazily provisioned container and the hour-long billable session. Which of the three you hold a reference to decides whether the second question still has the data.
Reality
Code execution and file creation is enabled by default on every Claude plan, and the same support article lists Team network egress as disabled in one section and on for package managers in another. Owners get four egress settings, and three of them allow outbound traffic.
Reality
Part one of a three-part benchmark series says isolating code execution is close to free and the GPU still has room at 10 concurrent agents. The same runs disabled tool calling and held every request to a single turn.
Reality
Durable state and sandboxed execution now arrive as platform primitives, so a dev.to guide argues the remaining failure mode is undocumented organizational context. It leans on one arXiv paper and one example.
Reality
Vendo's open-source layer lets B2B customers generate features from a vendor's own APIs, acting as the signed-in user. The controls it ships are company-documented, not independently audited.
Reality
The open-source framework ships durable execution, sandboxing, approvals, subagents and evals in public preview. If the bet lands, your in-house harness is now maintenance.
Reality
The Agents SDK now carries durable execution, sandboxed code execution and a durable filesystem. Cloudflare's argument is that harnesses cannot own those problems, which makes them a platform decision.
Reality