Should the older Phase 1 hard gate still be authoritative, or has later macro/data-proof work superseded it?
CSM decisions
Answer the unknowns without hunting through Markdown.
Each box is pulled from the canonical source-coverage file. Save short answers, leave things open deliberately, then tell Codex to apply the answers back into the planning docs.
Is the CM-owned vs CSM-owned client split from cm_csm_adac_aspec_product_vision.md still authoritative as of 2026-06-23?
Are <Client>.mapfile, <Client>_field_code_list.pdf, <Client>_attributes_breakdown.pdf, and <Client>_schema_summary.pdf mandatory for every production delivery pack, or only where the client workflow needs them?
What is the exact A-SPEC boundary: field/schema/package support only, or any future GIS deliverable path?
Which tab model is canonical for the real macro: source Seeded Setup, Element Review, Project Values, Run Evidence, Settings; or the older canonical Seeded Groups, Project Values, Run Evidence, Help?
Should file boxes be referred to by names only, not file-box numbers, because source docs conflict on numbering?
Should the prototype shell in C:\Users\a.almouhajer\Downloads\CSM_Apply_Profile_UI_Shell_v2_prefilled_extracted be moved into a stable CSM repo path before future agents rely on it?
Should exact historical proof metrics such as 91 official calls, 1039 passed, and 82 PASS / 0 FAIL / 0 MISSING / 2 SKIPPED remain canonical, or should only current rerunnable commands be canonical?
Should REFERENCE_POINT_A, ATTMF_FAILED rc=..., and QML filteredEmptyMessage be hard requirements or changelog/source-evidence details?
Is *.12d_runtime.txt the locked operator-facing sidecar convention?
Is the exact source_kind enum a hard runtime contract now?
Should field_visible defaults by source_kind be accepted as production field-export rules?
Should exact proof paths and counts from gap/audit docs be copied into canonical docs, or kept as linked source evidence?
What is the real MRBC_CADASTRAL_LOTS_MODEL 12d model name?
Which source-backed match_attr or att_key qualifier is approved for each of the eight duplicate Veris-code groups?
Should documented human approval be allowed to unblock a duplicate bridge without a source-backed discriminator, or must the approval specify the exact discriminator field/value?
What is the accepted MRBC production ADAC version/capture guideline for the real job?
Where is the ADAC 4.2 Data Dictionary XLS, or should choice proof remain formally blocked?
Should ASLID -> SWLID become a Veris master-code addition, a client alias/extension, or a rejected mapping?
Does the later WAADAC ADWHD bulk comment write/readback proof supersede the source research brief's Comments missing failure, or are they separate proof runs?
Are surface/terrain/cadastre assets in scope for the current WAADAC/MRBC acceptance path or only future ADAC clients?
Is G7 closed because deployed pages return 200 OK, or still open until mobile/browser screenshot and comment save/load proof are rerun?
Should status-truth PASS ever be green in UI, or always amber/qualified as "consistency only, not product completion"?