Status: root author and separate Music-agent desk review, 30 September 2026. These are proposed teacher/parent workflows, not child, educator, family or funder trial results.
Walkthrough and changes
Teacher: chose one useful puppet-book job and checked the actual Guide A facts/source against Digital Technologies W9–10. The separate agent found that locating an entry alone did not deliver the representation goal; the worked card now explicitly has the child place/draw/direct a picture/word representation with its given source/location. Target is partial K02 only; device use and personal-data identification are separate. One-mark art uses actual drawing/director-supported making, not a spoken future plan counted as finished art.
Parent/carer: followed the three-minute drawing invitation with familiar materials or adult direction. The separate agent flagged field burden, short writing lines and unexplained AAC. Cards now say to fill only useful fields, home wording spells out communication board/device, and wider lines give handwriting space. The teacher card adds a second writing line. Keeping, stopping, an unnamed picture and another short sitting are valid choices; no completed form or homework return is required.
Learner: the proposed invitations allow a child to choose a mark, placement and meaning or direct an adult. These are adult desk simulations, not observed child comprehension, enjoyment or independence. A supplied model remains separate from actual-room data; held check sources are not rehearsed here.
Funder/reviewer: source/version/criterion can be recorded, but the forms add no curriculum approval, whole-year certification or efficacy evidence. Original layouts, editable sources, licences, exact wording and checks are inspectable. School reporting and actual learner evidence remain local.
Actual artifact checks
- The installed LibreOffice document renderer opened both actual final Word files and produced two A4 teacher pages and one A4 home page in temporary storage. Root inspected all three rendered pages alongside the three shipped PDF pages. This is actual document rendering, not a human Word-interface trial or physical printing test.
- All card words are present in Word XML and shipped searchable PDFs. Word files are unprotected, macro-free and have no external document relationships. PDFs have measured page/text bounds and embedded fonts for all displayed text. Shipped PDFs are untagged; plain Markdown and editable Word alternatives retain the complete wording.
- Four generated files reproduce byte for byte in temporary storage. Fixed build/ZIP metadata clocks are reproducibility values, not learner/source-event dates. Generator requires Python 3, python-docx, reportlab and DejaVu fonts; audit additionally requires pypdf and Poppler inspection tools. From this folder use
python3 verify_tools.py; intentional manifest refresh is--write-manifest. Intentional generation usespython3 generate_cards.py --output-directory .. - Three temporary-copy corruption probes were rejected: removed Word field wording, document protection and an external document relationship. Originals were not mutated by these probes. Twenty-one local links/heading anchors and canonical worked source facts are checked. Default audit preserves pack bytes and modification times; file hashes exclude disposable caches.
- Separate Music-agent review independently checked the source facts, target, practical invitations and earlier actual PDF layouts. Root resolved its four concrete improvements; it did not claim to open Word in a human UI. Formatter/lint and the final read-only audit passed.
Remaining: actual educator/parent use and preparation time, local programme/state fit, individual screen-reader/AAC/tactile access, physical printing and learner feedback. No learner record or third-party submission is generated by the tools. This small guide helps adaptation; it does not complete the Foundation year.