The 16-slide runtime guide and PDF now correct five existing slides and add three adjacent explanations. The changes separate wire/runtime validation, show all eight segment fields, distinguish two envelope records, explain the tap priority branch, expose the first-slot startup gap and describe reconnect and backpressure precisely. Purpose-specific headings, equal peer emphasis, record specimens, a decision diagram and a three-slot state table replace compressed or misleading explanations.
Eight current recordings provide 12.6 minutes of regenerated narration. The library contains 79 guides, 1,028 unique slides, 1,051 guide-slide instances, 493 cited source entries and 1,028 recordings. All 1,020 unchanged canonical slides, scripts, manifest entries and recording bytes are preserved. No complete handbook unit is newly assigned; citation count is not teaching acceptance.
Implementation baseline: c57a3bbffebff018995a96466a8073e4f3f6772a.
Declared source pin: 1ceae52c7896a59ac74a4429d410f3218f0b91c6.
The authoring brief records
scope and slide jobs; the change manifest
binds the five reviewed before/after fingerprints. The legacy Conductor graph
is replaced because it implied a streaming first slot that initialization does
not establish.
- 22 executed source characterizations and 51 unmodified owning tests pass. The exact source modules and dependencies are used. Transpilation is not full application typechecking. BFF normalization is a source-reviewed excerpt; no HTTP route, GPU, deployed player or native device is executed.
- The final guide review passes 116 checks with 40 captures: four standalone widths, three embedded layouts, accessibility automation, four enlarged-text cases and two rejected negative controls. The full affected guide is checked, including eight unchanged slides around the correction sequence.
- Presentation Center checks navigation, source/transcript access, audio boundaries and the 16-page PDF. The 78 other guide HTML files and shared reader are preserved against the baseline.
- All PDF page text/order and word bounds pass. The agent inspects all 16 final pages and records the optical review. Legacy numbered peer columns on the unchanged overview, blend, compose and editor slides remain a tracked migration item in the optical record. Full guide design acceptance stays open. The draft review found a cramped response field label and awkward whitespace-case wording; both were fixed.
- Focused presentation tests, encoded audio/fingerprint checks, builder and coverage freshness, ESLint, Ruff, Prettier and patch whitespace pass. Exact commands and process exits are retained in the logs and command record.
The initial source fixture lacked the host alias for a real noble dependency; adding that resolution allowed the unchanged owning suite to run. The initial narration launch used system Python without soundfile, and the first Center PDF check inherited Python 3.9 without strict zip support. The existing Python 3.11 Kokoro environment supplies both requirements. Those setup failures are retained in logs. Neither required a product-source change or a weakened assertion. Two browser runs were stopped while diagnosing slow shutdown. Progress logging localized the delay after all assertions, and a complete run then exited successfully. The guide review tool now bounds element/navigation waits and reports its final phases; its successful full guide run is retained. Browser shutdown itself is not claimed to have a new deadline.
Human normal-speed listening, representative-reader acceptance, native behavior and a complete live Living Scene remain unperformed. This receipt closes the technical correction delivery only. The full V1 audit and wider Living Scenes review remain active. Existing historical receipts are preserved and describe their original revisions.
Use Python 3.11 or newer with the presentation dependencies. Run
python3 docs/presentations/presentation-center/verification/v1-living-scenes-corrections-2026-09-09/verify-receipt.py
at this package's publication revision to verify the frozen inputs, outputs,
source continuity and all recording hashes. Publication uses normal hooks and
origin main; later legitimate content changes may invalidate this historical
receipt when run at a newer revision.