# Author desk run-through — scope and limits

**30 September 2026 · version0.1.0-draft.** This is an author reading/artefact/browser simulation from teacher, parent, learner and funder perspectives. No child, family, teacher interview, classroom pilot, physical paper fit trial, audience consent or external accessibility review was performed. The observed evidence below concerns authored files and rendered outputs only. Remaining human gates are explicit.

## Teacher pass

Read every ten-day script and its A/B/C routes, prepared materials, twenty worked choices, check/key pair and integrated W19/W20 sources. Each script has six phases 3+4+5+7+4+2=25 minutes; actual sessions/check components may pause/continue later. Outcome moves from one broad crossing to one small trial, an original panel, selected dated stages and an original three-picture maker card. Adult cutting/fixing/lifting and child art direction are separately recorded; no compulsory fine threading, touching, speaking, writing, family materials or public sharing. Teacher reads one prompt/source chunk at a time.

Desk scenario: author follows Day92 with an imagined snag. Script stops/adjusts safely, records the particular event or NOT TRIED, and keeps child keep/change decision. This scenario is invented for desk review, not an actual paper result. Front means horizontal strip covers vertical band; behind reverses cover order. A flat segmented collage/drawing can communicate the plan while physical interlacing stays unmade. Made drawing/card and depicted unmade panel have different statuses. No universal material/strength, movement, engineering, cultural-tradition or source-place conclusion follows.

Teacher still needs to prepare/check actual broad bands and slack, agree local adhesive/tools/desk safety, review individual access, supply actual dated child-owned samples or labelled fictional substitution, and document actual responses/support. An actual optional maker/work encounter with stated particular purpose is required before related E01 evidence; community/context breadth is not supplied by generic diagrams. Actual willing sharing is required before P01. These are prospective human gates, not author-verified events.

## Learner desk pass

Author read child prompts and tested the adult-operated choices. “Want a strip to hide or pop out?”, “keep plain or add your own mark?” and “what job should your card do?” invite actual art decisions. There is no winner, fixed meaning, neatness score or forced finished grid. Child can keep a private original panel/card, make a drawing/flat collage, direct an adult or preserve an unmade plan. The supplied screen offers only one original strip/curve plan; child’s paper art can differ. Keeping plain or keeping privately are substantial choices.

Potential load remains: five screen choices plus a full exact state description can be too much to present at once; the teacher/guide explicitly reads one control and one source crossing at a time. On narrow screens controls and preview stack, so a child may need an adult to scroll between them. A full paragraph is an adult description, not a reading requirement. The three-picture card is approached one blank field at a time. Child engagement, comprehension, access and physical setup require a real optional pilot; they were not inferred from author simulation.

Fresh AA uses two horizontal bands and a separate vertical strip, unworked. Fresh AB uses a new two-band fictional dated sequence, not actual child stages. First response precedes key/model feedback, including four check-day swap cells labelled AFTER first check capture only. Source alternatives give facts without desired original design. AA front/behind transfer and AB bounded change/own guide decisions are not rehearsed in the interactive. Author checked fixed answers against actual rendered source: AB added strip on5May, added lower-right curve on6May, same front-left/behind-right crossings. Actual purpose, strength and audience remain unknown. A short follow-up may finish either check without speed judgement.

## Parent/carer pass

Read all ten optional0–3minute bridges. Each can be declined; none needs purchases, scissors, prints, device, upload, evidence return, teaching family or a willing audience. A home show-and-tell requires both people wanting it; privacy is a valid choice. Parents can distinguish an actual drawing from the unmade paper construction it depicts and do not need to validate material success or curriculum achievement. No child-specific diagnosis, home information or names/photos are requested.

## Funder and source pass

Four exact Foundation Visual Arts rows/descriptions and eleven integrated-context rows are pinned to the official workbook. Actual workbook cells, imported records and snapshot corroborate code/description; fresh HTTP200 workbook matches byte SHA. Three independent QCAA HTTPS requests returned403. Official alignment indexed text and complete three-page techniques indexed wording were observed; PDF screenshots failed, current bytes/layout unverified. School still confirms current local documents/assessment programme. Museum primary process reference supports general over/under paper crossings; all diagrams/progression/checks are original, with no borrowed motifs/images/cultural narrative or efficacy claim.

Original educational text/art CC BY4.0; implementation Apache2.0; DejaVu font separate. Curriculum, external work, later child art and community-held knowledge retain applicable rights. Actual local authority, reproduction/display permission and consent before any optional outside encounter/use remain held. Generic crossing art cannot establish a named cultural tradition or Country/Place knowledge. No pilot or whole-standard/term-grade achievement is claimed.

## Print and browser inspection

Generated all eight SVG/searchable PDF pairs. Each actual PDF was rendered; [contact sheet](print/previews/contact-sheet.png) inspected across all eight pages, with full-size crossing-source and fresh AB renders additionally inspected. Page geometry is A4 portrait and all extracted word bounds lie within the page. No clipping or prefilled child response found; front/behind covers visibly match wording. Exact visible wording/order, spatial descriptions, optional safe tactile/no-print/no-touch routes are supplied in [linear alternatives](print/TEXT-ALTERNATIVES.md). PDFs are not tagged; linear alternatives remain the reading copy. These observations do not validate a physical print’s contrast, accessible fit or child use.

Local installed Playwright exercised144 cases (all8 crossing orders×2marks×3uses×1280/390/320), verifying actual SVG cover nodes against exact description/ARIA, all-same/mixed process distinction, curve/source positions, native selected option fit, SVG bounds/global overflow, keyboard focus/change/reset and reload. No page errors or outside requests. Author inspected [desktop](digital/previews/desktop.png) and [390px mobile](digital/previews/mobile.png) full-page previews. BrowserAct skill/runtime had no configured API/browser; local Playwright was the documented fallback. No BrowserAct run is claimed. The planner code contains no network/storage API or persistence and checks/key are not loaded. Author captures are not child trials.

## Release posture

Default verifier and browser QA are read-only. The manifest covers authored docs/source pins/code/licences/all sixteen A4 assets and previews; it excludes itself. Explicit `--write-manifest`/`--capture` switches are author writes. Final frozen verification checks content/hash consistency and deterministic regeneration; real school/child/accessibility/cultural/audience gates remain above. Files are uncommitted by this author, with no shared map edits, staging or push.