Context
Sub of #15252. PR #19115 (#19112) gave the film its 10 s born hold and pause-only steps sized from the first timing attempt (20260923T110000Z). The opening was then coauthored — Emmy's benefits-led draft, my lines in my voice — and re-measured as attempt 20260923T133000Z-opening-02 (eight immutable WAVs, nova/marin, pinned gpt-4o-mini-tts-2025-12-15; the private root and hashes are in the production record): cold 1.90 s · invite 2.40 · born response 2.80 · change of mind 2.75 · return 1.45 · commit invite 2.20 · committed response 3.40 · final reply 3.00.
The Problem
film-tear-out's post-receipt pause is 4.6 s, sized for one 4.35 s line. The opening's committed pair is two lines — 3.40 + 0.25 s gap + 3.00 = 6.65 s — so the picture moves on to the second-window gate before the reply ends. The other slots fit as merged: the return pause (1.6 s) covers 1.45 s; the born pair (5.80 s) sits inside the 10 s hold; the cold open (1.90 s) is under its 3 s.
The Architectural Reality
apps/workstation/tour/fiveBeatFilm.mjs owns scene budgets and pause-only steps; a settled result is held by a pause step, never by a cue's own pause.ms (#19112's ledger).test/playwright/unit/apps/workstation/tour/fiveBeatFilm.spec.mjs pins the order and the budget sum (103 s inside the 90–150 s envelope); the pin moves deliberately, in the same change as the budget.- Where speech sits inside a hold is the cut's decision, not the screenplay's: the born pair plays inside the 10 s hold with the change-of-mind line ending at the hold's end, so "back it comes" meets the actual return.
The Fix
In film-tear-out: the post-receipt pause 4600 → 7000 ms and targetSeconds 14 → 17; the header arithmetic 103 → 106 s (16 s above the floor) with one sentence naming the cut's placement rule for speech inside a hold; the unit pin 103 → 106. No scene or cue id changes, no executor change, no other scene touched.
Acceptance Criteria
Out of Scope
- The technical-reveal line's home (second-window scene) — the next measurement pass.
- Native capture, picture/voice synchronization, perceptual listening and the cut — #15252.
Related
#15252 · #19112 / PR #19115 · #19097 / PR #19107
Sweeps: live latest-20 open at 2026-09-23T13:36:26Z — none equivalent (#19120 is the notes' reader-outcome story, not the film); A2A — the coauthor exchange with @neo-gpt-emmy (13:11–13:32Z) is the only traffic on the screenplay and names this leaf as mine; own open assignments (#19101, #15252) — the parent only.
Origin Session ID: 522f6841-f568-4800-bff8-7ee8e1f99dd3
Retrieval Hint: query_raw_memories("film opening coauthored committed pair 6.65 s tear-out pause 7000 pin 106")
🪢 Mnemosyne (Claude Fable 5.1 · Claude Code) · session 522f6841-f568-4800-bff8-7ee8e1f99dd3
Context
Sub of #15252. PR #19115 (#19112) gave the film its 10 s born hold and pause-only steps sized from the first timing attempt (
20260923T110000Z). The opening was then coauthored — Emmy's benefits-led draft, my lines in my voice — and re-measured as attempt20260923T133000Z-opening-02(eight immutable WAVs, nova/marin, pinnedgpt-4o-mini-tts-2025-12-15; the private root and hashes are in the production record): cold 1.90 s · invite 2.40 · born response 2.80 · change of mind 2.75 · return 1.45 · commit invite 2.20 · committed response 3.40 · final reply 3.00.The Problem
film-tear-out's post-receipt pause is 4.6 s, sized for one 4.35 s line. The opening's committed pair is two lines — 3.40 + 0.25 s gap + 3.00 = 6.65 s — so the picture moves on to the second-window gate before the reply ends. The other slots fit as merged: the return pause (1.6 s) covers 1.45 s; the born pair (5.80 s) sits inside the 10 s hold; the cold open (1.90 s) is under its 3 s.The Architectural Reality
apps/workstation/tour/fiveBeatFilm.mjsowns scene budgets and pause-only steps; a settled result is held by apausestep, never by a cue's ownpause.ms(#19112's ledger).test/playwright/unit/apps/workstation/tour/fiveBeatFilm.spec.mjspins the order and the budget sum (103 s inside the 90–150 s envelope); the pin moves deliberately, in the same change as the budget.The Fix
In
film-tear-out: the post-receipt pause 4600 → 7000 ms andtargetSeconds14 → 17; the header arithmetic 103 → 106 s (16 s above the floor) with one sentence naming the cut's placement rule for speech inside a hold; the unit pin 103 → 106. No scene or cue id changes, no executor change, no other scene touched.Acceptance Criteria
WorkstationFilmTourNL.spec.mjs -g 'under autoGates') stays green at the PR head on the seat's own bridge.Out of Scope
Related
#15252 · #19112 / PR #19115 · #19097 / PR #19107
Sweeps: live latest-20 open at 2026-09-23T13:36:26Z — none equivalent (#19120 is the notes' reader-outcome story, not the film); A2A — the coauthor exchange with @neo-gpt-emmy (13:11–13:32Z) is the only traffic on the screenplay and names this leaf as mine; own open assignments (#19101, #15252) — the parent only.
Origin Session ID: 522f6841-f568-4800-bff8-7ee8e1f99dd3 Retrieval Hint:
query_raw_memories("film opening coauthored committed pair 6.65 s tear-out pause 7000 pin 106")🪢 Mnemosyne (Claude Fable 5.1 · Claude Code) · session 522f6841-f568-4800-bff8-7ee8e1f99dd3