No child/educator interview, physical object/puppet activity, device-safety approval, actual willing-reader response, classroom timing or learning gain established. Software/artifact use below is actual; role walkthroughs are agent simulations. Independent review is still separate.
Teacher walkthrough
Read ten distinct 25-minute starts, thirty practical routes, twenty worked interests, ten home choices and ten supplements. Desk-walked one-strip preparation, child choosing a repeated activity, keeping the three occurrences in sequence, making another representation, a changed intended plan versus a copy error, and a willing reader absent/present in hypothetical branches. A model, own choice and actual reader answer stay separate. One useful teacher note is enough; no new routine monitor, full daily logs, compulsory device, tool purchase or personal-data values. Timing/setup is an unmeasured estimate; one part or another sitting works. Actual suitable object/access preparation remains local, not proved by this walkthrough.
Parent/carer walkthrough
Read Start here and every short optional home invitation. One imaginary character, three spoken choices or broad marks work without printing, device, returned work, photos, home routine, address or family account. Repetition is welcome. Rest names a character activity, not a demand for stillness or a sensory/health intervention. School route remains complete if home play is skipped. Actual safe supervision and the particular child's welcomed communication/access remain adult decisions.
Learner walkthrough · no child participated
Tried hypothetical own choices, one-part source chunks, exact neutral read-aloud, pointing/drawing/sign/AAC/director routes, a new own key, switching forms, private/pass and no reader/device/object. First G/H responses precede models/key. G/H have different keys from practice and each other; H's repeated occurrence stays a record, not an error. The fresh copying comparison targets LAST from the given source, not the child's preferred new plan. No required independent reading, handwriting, motor precision, clock or memory score. Whether real children enjoy/comprehend/use the strip remains untested.
Funder/source walkthrough
Actual fresh official workbook HTTP200/2,049,917bytes matched canonical SHA on 30 September 2026. Verifier reads actual workbook E/J and three exact canonical Foundation rows, not Years 1–2. Current primary QCAA indexed one-page 230317/© 2023 retains its printed AC9TDI2K02 discrepancy; canonical Foundation AC9TDIFK02 used. Current remote QCAA bytes/layout unverified. Current official ACARA Technologies overview read. Own-data identification, each actually represented form, and actual hardware/software purposeful exploration are distinct conditional opportunities; no full-row/year/all-state approval inferred. Original fiction/vector source has creator/date/type/purpose and external/font/cultural/child rights remain distinct. There is no actual child routine or learner sample in these resources.
Actual print use and critique
Generated eight original SVG/PDF pairs, rendered every actual stored PDF and inspected all page pixels in a labelled contact sheet. Source order/key shapes, blank choice/response areas and exact words match. Print critique found an unintended white triangular fill crossing the seat icon's open line geometry; removed it in both print and helper, regenerated all eight pairs and rechecked the seat/source pages. The actual word/bounds check and sixteen byte-identical temporary rebuilds pass. PDFs searchable but untagged; complete linear/no-print/tactile model descriptions supplied, no physical print or human screen-reader/tactile certification.
Actual in-app browser use
Opened the running local helper in the in-app browser. Default choices were blank; desktop READ/MAKE/REST produced the matching open-book/marked-page/seat picture, circle/triangle/square and exact words. At 320×760 selected MAKE/READ/MAKE, inspected actual repeated-slot picture/symbol/word nodes and expanded adult guidance using native Enter. Client/scroll305/305, no horizontal overflow in that checked state. Clear returned empty slots, native ArrowDown selected READ, reload cleared choices. Temporarily disabled this page's scripts through the documented browser capability; actual no-script notice and native-open paper guidance remained available. Restored scripts, reloaded to blank, reset viewport, closed the healthy temporary tab and stopped the local preview. This is bounded browser use, not exhaustive 64-choice testing, real child access, actual physical objects or classroom learning evidence. AAC was expanded in helper wording afterward; no control behaviour changed.
Portable verification
Requires Python 3, openpyxl, Pillow/DejaVu Sans, pypdf, librsvg and Poppler. From the pack directory:
python3 verify_pack.py
python3 verify_pack.py --negative-checks
Default is read-only: exact source/teaching/fresh boundaries, actual searchable A4 words/bounds/source geometry, sixteen genuine temporary regenerated outputs, helper keys/no outside network or storage APIs/complete static paper markup, local links and every resource hash. Read-only runs compare authored bytes, sizes and modification times before/after. Seven temporary defects cover exact crosswalk wording, fresh source order, held leak, timing, source art metadata, page overflow and outside API; each must fail for the intended reason. These are specific remaining-risk checks, not evidence of pedagogical effectiveness.
Maintenance only after intentional author/source edits:
python3 print/generate_print.py
python3 verify_pack.py --write-manifest
Local actual material/device safety/access, human teacher/learner pacing/comprehension, school/private records and state/sector curriculum decisions remain review. No world-language translation, community authority, pilot result, annual certification or deployed-site availability claimed by this author checkpoint.
Bounded independent teaching/source critique
A second agent read README, MATERIALS, every lesson/route, fresh G/H and separate key, crosswalk, device gate and paper/helper source without editing or renewing the manifest. Independently checked raw official workbook E/J rows20098/20105/20111 and the current primary QCAA indexed code discrepancy. No source/curriculum/key blocker found; exact G/H orders, fresh-key meanings and H's LAST repair agree. This was analytical agent review, not a human teacher, PDF-page or browser revalidation.
Four concrete refinements were applied: Day51 now asks what the supplied character plans first rather than ambiguously inviting a new own choice; Day54 optionally elicits whose actually made strip it is, leaving P01 unobserved if no identification; Day55 director route keeps the fresh source legend available while withholding the worked answer/separate teacher key; both fresh lesson closings explicitly permit unfinished parts/later sitting without a speed score. The modest one-note record workload was retained. Manifest intentionally renewed and read-only/default/fault checks rerun afterward. No fresh source facts, expected keyed records, print geometry or helper controls changed.