My project assessment agreement

Public-safe requirements only: fictional data and aliases, no grades, real identities or private coaching. See the rubric. Draft scope at Session 02; complete observable cases during Session 04. This agreement describes the existing project assessment, not a separate test.

Product / audience / problem:
Standard: AIQAA-1.0
Agreement version and exact commit (in the review record):
Three connected core features and non-goals:
Main journey and required views:
Record types, relationships, transitions and persistence promises:
Synthetic seed/reset procedure and expected state:
Supported OS/browser/runtime versions; local run/check commands:
Review matrix: 375 / 768 / 1440 CSS px, 200% zoom and 320 CSS px reflow, keyboard journey; approved format equivalents:
Integration boundary (local file exchange is enough), recovery rules and repeated-action behavior:
Three highest-priority risks and why; expected mitigations:
Design sheet: type, spacing, colors, navigation, action hierarchy and justified component exceptions:
UI-state inventory by view: empty / populated / invalid / success / recovery / waiting where applicable:
Review milestones and evidence index location:
Budget and no-cost fallback:

Core acceptance cases

Use stable case IDs. List every promised core behavior and each relevant negative/boundary outcome. “Fast,” “pretty” and “no bugs” need observable definitions. Do not use arbitrary limits simply to make a case measurable.

Case ID / feature / risk	Initial synthetic state and environment	Exact input/actions	Expected visible result AND saved state	Reset / evidence location

Include the complete main journey, persistence, recovery, invalid/malformed data, repeated action, meaningful relationships/calculations, responsive/keyboard behavior and all promised core functionality. Define expected behavior before running the case. No provider accounts or paid services are necessary.

Criterion-to-project map

All rows are required; map equivalent evidence, not N/A. IDs are defined in the rubric. The same artifact can cover several rows.

Criterion	Project-specific cases / artifacts / observed behavior	Equivalent route, if needed, and why it demonstrates the same skill
P1
P2
P3
V1
V2
V3
Q1
Q2
Q3
H1
H2
H3
G1
G2
G3
A1
A2
A3
B1
B2
B3

Change log

Version	What changed and why	Affected cases / criteria	Equivalent evidence and implementation/review plan

The instructor keeps approval identity/date, agreed equivalents, accommodations and scope-change decisions privately, tied to this version and SHA. Keep past versions. A failed core case cannot simply disappear at final review. Unresolved expected behavior is a clarification item, not permission for AI to invent a requirement.