Client evidence for feature and schema decisions

HVE-DECISIONS-1.0. All interviews, proposals, changes and review findings are fictional.

Use these with ZAVA-PP-2.0 policy/examples. They replace the earlier implementation/property/debugging baselines in the core.

How to read a case

F facts: client interview evidence, constraints and unanswered needs.
P proposals: candidate choices to examine, not an approved design.
C changes: new requirements whose scope/data consequences need a human decision.
T findings: supplied walkthrough feedback on a candidate model, not executed application tests.

Original policy remains authoritative. A future request is not automatically accepted scope. CSV examples are not a normalized schema. Do not inherit a table list from earlier maker material or treat more entities as a better answer.

Case	Decision problem
01 CoE	Which delivery capabilities/service features and use-case/decision/activity data are necessary?
02 Orders	Which extraction/review/correction features and source/order/line/history relationships are needed?
03 Process	Which insight/intervention features need event, current-work and follow-up data?
04 Canvas inspection	Do offline, photos, checklists and remediation justify particular records/relationships/history?
05 Suppliers	How do self-service, contacts/legal entities, correction and private review affect the model?
06 Agent	What should the agent know/do and what is transient versus persistent business data?
07 Legacy service	What operating features and job/attempt/observation/resolution data support safe decisions?
08 Dispatch	Which interaction, skill, assignment and accountability features justify the app/model?
09 Value	Which reporting/casework features need source grain, lineage and investigation history?
10 Reuse	Which features/schema choices are standard and which need client-specific variation?

The agent may see all sections. These are learning cases, not blind evaluations; facilitator holdouts remain separate.