Module dated_context 

Format recalled facts as a chronological, date-prefixed timeline with a “now” anchor — the dated-context representation measured to lift temporal question answering, shipped as product behavior rather than a harness prompt. Dated context: turn recalled facts into a chronological, date-prefixed timeline with a “now” anchor — the representation measured to lift temporal question answering (the examples/locomo temporal ablation: +33.6pp, McNemar p=1.8e-28). This ships that representation as product behavior so a caller reproduces it through the installed API instead of re-implementing the formatting in a prompt.

The date lives in caller-supplied metadata under a field the caller names (e.g. ts, occurred_at), holding a YYYYMMDD integer — the same key shape recall_where filters on. A fact whose named field is missing or not a valid YYYYMMDD date is treated as undated: it still appears, just without a date prefix and after the dated timeline, so an unlabeled fact never invents a misleading date.

Only the formatting ships here — not the LLM reasoning prompt the harness wraps around it. Presenting retrieved facts is a memory-store concern; prompt engineering is the caller’s.

Structs

DatedContext

A chronological, date-prefixed rendering of recalled facts.

Functions

format_dated_context

Render facts as a DatedContext, reading each fact’s date from the date_field metadata key (a YYYYMMDD integer).