The rules that fired, queryable from your stack.

oddly runs a fixed library of deterministic playbooks against an account: which rule applied, when, to what, and what it recommended. oddly Playbooks exposes that run history and rule coverage as a programmatic primitive, so a system can see what oddly has already evaluated rather than re-deriving it.

The idea

oddly's playbooks are deterministic by design: the same account state always triggers the same rule. That makes their output safe to expose. A partner can ask which playbooks ran for a brand in a window, which rules covered which surfaces, and what each run concluded, then build on top without guessing at oddly's internal state.

Where it sits

• Run history: the playbooks that evaluated an account and what each concluded.
• Rule coverage: which rules are watching which surfaces, so gaps are visible.
• Deterministic and reproducible: a run reads the same on every query, with no model drift.