# For agents: What the author controls

The ideal OpenProse · board 2 of 17 · 2026-10-09.
This deck describes a proposed experience.

Keep meaningful requirements stable as the model, agent, and tools change.

## On the slide

**What must hold can outlast how it gets done.**

A model changes.

Its explanation must remain accurate, supported, and reviewable.

## Explanation

Consider a fictional quantitative forecasting model. Today someone documents its assumptions, calculations, and limitations. Tomorrow the model changes. A better agent or a different tool might do the next update. The author should be able to retain the same meaningful requirements while execution improves. This is the contract-engineering thesis. Prompting, context selection, and harness design remain important work inside the system. The ambition is that authors can increasingly direct capable systems through the requirements that endure, instead of repeatedly reconstructing every method.

## Whiteboard sequence

Open only the frames useful to the question. Later frames add to the same board;
the final frame contains the complete drawing.

- [Frame 1 — black core](https://prose.md/visual-guide/2026-10-09/ideal/slide-02-layer-01.png)
- [Frame 2 — cumulative reveal](https://prose.md/visual-guide/2026-10-09/ideal/slide-02-layer-02.png)
- [Frame 3 — final](https://prose.md/visual-guide/2026-10-09/ideal/slide-02-layer-03.png)

These explanations add no normative requirements. Read the
[selected kernel 0.3.0-rc.3](https://pkg.prose.md/releases/0.3.0-rc.3/core/README.md) for language rules.

[Previous: Contract engineering](https://prose.md/visual-guide/ideal/01.md) · [Next: A readable agreement](https://prose.md/visual-guide/ideal/03.md)

[All The ideal OpenProse boards](https://prose.md/visual-guide/ideal.md) · [Guide index](https://prose.md/visual-guide.md) · [Agent entry](https://prose.md/agents.md)