OpenAI Codex Made Crystal Clear · chapter 2: One Agent, Four Surfaces
Harness anatomy
2026-10-08
System prompts (including your AGENTS.md) and managed context feed the model; the model drives its tools inside a sandbox that limits what is physically possible; approvals sit at the edge deciding when you get asked; checkpoints snapshot the repo around every task. Chapters 4, 10, and 13 each take one of these layers deep.
Below: the paragraph from the book that builds this idea, then the diagram itself (Figure 2.3), and a recap. About a minute of reading.
One more zoom before the vocabulary. Figure 2.1 drew the harness as a thin dashed ring; Figure 2.3 cuts one surface open (the CLI, since it is the one you will meet first) and names every layer between you and the model. You will spend chapters 4, 10, and 13 turning these layers to your advantage, so meet them once now, gently.
Recap
- The idea: System prompts (including your AGENTS.md) and managed context feed the model; the model drives its tools inside a sandbox that limits what is physically possible; approvals sit at the edge deciding when you get asked; checkpoints snapshot the repo around every task.
- The picture: Figure 2.3, from chapter 2 ("One Agent, Four Surfaces") of OpenAI Codex Made Crystal Clear.
- Go deeper: the chapter builds this step by step, with recipes and sources at the end.
This diagram is one of many in OpenAI Codex Made Crystal Clear.
Every chapter opens with the gist, draws the hard ideas, and ends with recipes and sources.
Get the book
