Teacher desk run: I can begin Day 121 with the supplied F text and three-stills print, then show G's visibly different solid/dashed routes. I can deliver ten 25-minute sessions using the six timed steps without inventing a physical outcome. Before Days 124/128 I must decide whether the exact real setup is approved, safe and accessible. If not, I write REAL OBSERVATION NOT AVAILABLE and teach the complete fiction route. I do not lift a textbook work sample, copy a child photo, improvise a ramp or call a hand-chosen path a material effect. I keep Day 125/130 cases unseen until first response.
Child desk run: In F I see a square at START, MIDDLE, FINISH. In G I can follow a straight solid line or an upper-bend dashed line; both fit the same three stills, so I can say “I don't know the path from F.” H tells me a pretend stop and turn but not whether the edge touched a hidden line. I can say, sign, point, use AAC, hear an exact source read or direct an adult. I need not touch a pad, move my body, look at bright media, speak publicly or discuss my family's things. “I don't know yet” is a useful answer. The family page costs nothing and does not need returning.
Assessor desk run: AB's orange rectangle/LOW LEFT–HIGH CENTRE–LOW RIGHT and AC's GREEN SQUARE 73/screen-hidden interval do not appear in F/G/H/J, learner routes, worked swaps or illustrated prints before first response. The checks test source-bound interpretation and a safe next question; they are paper-only. I log access, exact first words and neutral reread before consulting the separate key. A paper source does not earn actual AC9SFI02 investigation or causal movement evidence. An attributed adult account does not become the child's direct sense. A genuinely earlier child prediction is compared only with a genuine later observation if both exist. A single check does not certify term/year mastery or growth.
Potential funder desk run: The pack supplies ten scripts, thirty routes, twenty worked swaps, three narration-ready transcripts, ten family choices, two fresh checks with keys, six illustrated vector/PDF pairs, exact text alternatives, local source provenance and a regenerable hash receipt. The delivery claim is inspectable; learner outcomes, accessibility certification, state-by-state endorsement and cost savings are not measured here. The next evidence step is human educator/accessibility review and a consented pilot with predeclared measures, not a claim that this draft is already 10/10 in classrooms.
Visual/print critique (30 September 2026): rendered all six PDFs and checked image layout, readable one-page text, actual square placement and solid/dashed route contrast. A first draft adult-hand mark looked like a three-pronged tool; the generator was revised to a closed hand outline and regenerated. The raw-attempt “NO FIXED RESULT” cue was moved above the footer. Verifier checks geometry and exact PDF text; it cannot judge child comprehension or braille quality. The six SVG title/desc strings and linear text alternatives are available for adaptation.
Next human checks: confirm local surface/apparatus handling, neurodivergent access and communicative equivalents with educators and affected learners; inspect printed A4 at classroom scale; pilot the case prompts without teaching to them; review the chosen state's current syllabus/assessment policy. None of these checks is represented as completed.