Shop the Official Harley-Davidson site for new & used motorcycles, genuine parts & apparel. Locate a dealer or take a test ride. Free Shipping +$50 for members.
Shop the Largest selection of Harley-Davidson parts & accessories at Dennis Kirk! Get your Harley Parts Fast with Free Same Day Shipping on orders over $89!
Harness engineering: leveraging Codex in an agent-first world By Ryan Lopopolo, Member of the Technical Staff Listen to article
OpenAI updates the Agents SDK with native sandbox execution and a model-native harness, helping developers build secure, long-running agents across files and tools.
The harness is the control plane around the model: it owns the agent loop, model calls, tool routing, handoffs, approvals, tracing, recovery, and run state. Compute is the sandbox execution plane where model-directed work reads and writes files, runs commands, installs dependencies, uses mounted storage, exposes ports, and snapshots state.
To kick off, we’ll focus on the agent loop, which is the core logic in Codex CLI that is responsible for orchestrating the interaction between the user, the model, and the tools the model invokes to perform meaningful software work. We hope this post gives you a good view into the role our agent (or “harness”) plays in making use of an LLM.
Claude Opus and GPT-5.4 xHigh, for me, are basically tied as models. On hard tasks, when one can’t handle it, I switch to the other and it usually solves it. Head-to-head, they’re both strong. What separates them today is the harness, not the model. And Claude Code’s harness is simply superior right now, for two concrete reasons: planning and parallel execution. Planning. Claude Code ...