# Living Scenes technique catalog and compatibility

The [technique catalog](../../decks/living-scenes-technique-catalog.html) teaches
all twelve V1 editorial techniques, all nine Blend Kernel primitive contracts,
the exact launch thresholds, the five-field continuity scorecard, eight
technique-specific observation shapes, all reduced-motion substitutions and the
actual versioning boundary. The connected
[compatibility guide](../../decks/living-scenes-compatibility.html) follows one
sky-to-lesson pair through four hard policies, seven equal-weight dimensions,
strict threshold regions, the complete twelve-by-five template matrix, tone and
context admission, candidate filtering and the bounded override record.

The forty-four slides sit between the member composition workflow and Compose
Assist. Exact tables retain closed catalogs, matrices and numeric bounds;
diagrams show transformations, independent measurement inputs, policy branches
and execution order; two worked cases expose all scorecard and compatibility
arithmetic. No slide uses positional emphasis or a generic `Question` evidence
heading. The library now has 83 guides, 1,103 unique slides, 1,126 guide-slide
instances and 507 cited source paths. The new narration totals 2,989.647 seconds
(49.8 minutes), and all 1,103 recordings verify against their current scripts.

| Evidence | Scope |
| --- | --- |
| `source/characterization.json`, `source/owning-tests.json` | Sixteen focused source cases plus all 77 tests in the six owning suites. Fourteen source and test files match the immutable source pin, the presentation baseline and integrated `HEAD`. |
| `browser/report.json`, `browser/focused-report.json` | 314 standalone/embedded checks with 176 captures, four rejected negative controls and six enlarged-text cases; a separate 132-check desktop, 390-pixel and 320-pixel pass with 88 captures. Zero browser errors, fit findings, accessibility violations or positional-emphasis findings. |
| `center/result.json` | All 83 guide routes and navigation behavior; 81 prior guide bodies preserved; both new guides at desktop/mobile embed sizes, four audio-boundary checks, 88 embedded slide captures and complete PDF export. |
| `pdf-text.json`, `pdf-bounds.json`, `pdf-pages/`, `optical-review.json` | Exact authored visual text and ordered fields on all 44 PDF pages, page-bound checks, Poppler renders, complete contact sheets and selected full-size optical inspection. |
| `commands.json`, `logs/`, `attempts/` | Final passing commands and retained discovery failures that produced the three layout corrections and the Python interpreter correction. |
| `state.json`, `verify-receipt.py` | Exact artifact, evidence and recording hashes plus preservation of every slide and recording from the 1,059-slide presentation baseline. |

Eleven complete handbook units now have
[reviewed assignments](../../assignments/v1-living-scenes-techniques.json).
The [source map](../../authoring/v1-living-scenes-techniques-source-map.json),
[review configuration](../../authoring/v1-living-scenes-techniques-review-config.json)
and [per-slide ledger](../../authoring/v1-living-scenes-techniques-ledger.json)
bind the route, teaching job, visual treatment, source qualification and final
slide fingerprint.

The source review corrected several tempting overclaims. The catalog freeze is
shallow. The main technique gate receives no tone band and does not read the
catalog's `requires` or `forbids` declarations. Tara blocks three identifiers
without inspecting the supplied pair tone. The general scorecard, catalog
thresholds and technique-specific observations are three separate contracts,
with no reviewed adapter binding all three. Four catalog identifiers have no
technique-specific switch case. Enforcement-point strings are error labels;
the tests do not prove one-frame timing or an application caller. The version
helper proves equality only, and the override constructor proves neither valid
record identities nor durable persistence. Primitive validation does not prove
DSP, shader or renderer execution.

The visual checks found and resolved three concrete issues. A five-source
relationship map exceeded the 720-pixel desktop workspace, so its closed
editorial grouping became a compact two-column table. The technique-specific
gate and compatibility-dimension tables each left only six pixels between the
printed table and surrounding content; closely related rules now share a row,
with every value retained and measured print gaps above twelve pixels. A first
negative-control selector targeted SVG geometry outside the generic semantic
auditor and was replaced with four required-node removals. The first PDF attempt
also exposed the host's Python 3.9 incompatibility with `zip(strict=True)`; the
final center run placed the existing Python 3.11 environment first on `PATH`.

The in-app browser runtime named by the browser skill was not exposed in this
session. The repository Playwright harness launched the installed Chrome 152
binary for all final browser and PDF checks. Agent optical review was performed;
human comprehension sessions and normal-speed human listening were not.

The immutable product-source pin is
`1ceae52c7896a59ac74a4429d410f3218f0b91c6`. The earlier presentation baseline
is `5b6250a843d198c1dc453c680aba3c4a6805128a`; the unrelated Eve work integrated
before final verification is `56571f09b020502933611ea074b185bd003846d2`.
Full Score Editor, Tara venue editing, downstream rendering and measurement,
distribution/sharing, the remaining V1 workstreams and whole-library human
acceptance remain open.

At the package's owning revision, reproduce its artifact audit with:

```sh
python3 docs/presentations/presentation-center/verification/v1-living-scenes-techniques-2026-09-09/verify-receipt.py
```
