Explanation

These pages explain the choices behind Ribbit and its examples. They are useful when deciding how to design a workflow or interpret its results.

• When to chain models: where a small intermediate step can help, and how to compare it fairly.
• Why the examples retain originals: a walkthrough of the triage, context, and brief designs.
• How records and pipes compose: data boundaries, identity, metadata, and flow budgets.
• Execution and trust: what models, built-ins, extensions, and external harnesses can do.
• Model evidence: measured results and known failure modes.
• Product direction: the intended users, scope, and product requirements.

For commands you can follow now, use the how-to guides.