# Tara scheduling and reminders · delivered increment · 8 September 2026

This guide follows `tara-ritual-sessions` in the active
[V1 program](v1-completion.md). Readers should be able to explain when an
interrupted practice is proposed again, why its time moved, which reminders were
created and what still requires an owning delivery system. The implementation
baseline is `c0001c6c3f5f26db07f90bb30569d314579ffacf`; the primary source pin
remains `1ceae52c7896a59ac74a4429d410f3218f0b91c6`.

Visual thesis: the existing cream field guide, with explicit clock values,
decision branches and record specimens. Give the changed time a dominant visual
position; keep peer policies and outcomes equally weighted.

Content plan: an evening return moves into quiet hours; inspect its plan;
separate declared policy from actual time selection; resolve the local clock;
follow quiet hours, active days and reminder shifts; then inspect storage,
delivery queries and reschedule identity before resolving the opening case.

Interaction thesis: use the existing source/transcript reveals, keyboard
navigation and audio controls. All timing, branching and ownership explanations
must remain complete in static HTML and PDF. No shared renderer or product
runtime change is planned.

The [discovery record](v1-tara-scheduling-discovery.md) records the complete
scheduler, operation and owning-test reads. This increment also read the exact
contract scheduling/trigger schemas and domain trigger interfaces. An initial VM
probe executes the pinned scheduler with explicit inputs: Node 24.13.1, ICU 78.2
and timezone data 2025c. It confirms the evening shift, primary reminder
coalescence, Monday-only overnight behavior and the selected New York/Lord Howe
clock cases. These are runtime characterizations, not delivered notifications or
predictions about future timezone legislation.

Before promotion, retain meaningful source fixtures for continuity, timezone
resolution, quiet-hour boundaries, disabled reminders, dual clocks, storage
isolation, due-query filtering and reschedule identity. Review the rendered
draft and exact numerical examples. Then deliver narration, PDF, catalog
placement and affected center checks, inheriting unchanged evidence only through
exact-byte continuity. No source unit is accepted merely because it is cited.
The feature's combined ritual/steps/scheduling unit still needs the remaining
practice and assembly explanation before whole-unit assignment.

The optically reviewed manuscript is now in `content/32-tara-scheduling.json`;
its final center/PDF delivery checks pass; see the retained receipt below. This
guide does not close the player, practice, companion, mentor, workbench or full
V1 review. Human comprehension and normal-speed listening remain separate,
unperformed acceptance work.

## Source-characterization checkpoint

Sixteen immutable-source VM tests pass in
`tools/presentations/tests/pinned-tara-scheduling.test.mjs`. They execute the
complete scheduler and the complete AST-selected reschedule function, using the
actual scheduler dependency, controlled time/randomness and explicit storage.
Cases cover evening coalescence, all continuity branches, equality and future
activity, ignored trigger admission, full/half-hour clock transitions, dropped
resolution reasons, quiet-hour boundaries and separate zones, overnight active
days, disabled/follow-up behavior, dual clocks, cloning, unfiltered lifecycle
queries, reschedule parent references and propagated input/storage errors.
Targeted ESLint passes. The fifteen-slide draft has 2,317 note words and three
questions per slide. Its initial builder and draft review passed. All fifteen
tracks are generated; final center/PDF delivery checks pass.

The first fifteen-slide layout review passes desktop, compact desktop, mobile,
source/explanation access and PDF checks. Model optical review inspected every
compact desktop slide. It found duplicated arrow glyphs where authored connector
labels repeated the renderer's arrow, a split clock interval and long record
labels. Six slides received composition-only polish; their notes and source
claims are unchanged. The first captures and report are retained before the
settled draft review.

The full final layout matrix found a 320-pixel overflow in the record heading's
long type identifier. The heading now reads Schedule plan; the exact type
remains in the explanation. This is a composition-only correction and changes no
speech. All fifteen slides now pass the 111-check
standalone/embedded/enlarged-text matrix, with sixty retained captures and three
rejected negative controls. The final record-heading desktop and the timezone
mobile capture were optically reviewed. The prior 644 manuscript/narration
entries and MP3 bytes are preserved; fifteen new tracks add 1,336.191 seconds
(22.3 minutes).

The final
[delivery receipt](../verification/v1-tara-scheduling-2026-09-08/README.md)
records a fifteen-page PDF with exact visual-body text, current center
navigation and first/last-track boundaries, source fixtures and optical review.
Publication of this increment leaves the full V1 program active. No live
notification, production scheduling acceptance, normal-speed listening or human
reader review is claimed.
