skills/behavioral-evaluation/references/case-index.md

# Case index

Provide the selected fixture packet to the execution task. Keep the separate assessor rubric out of that task. All fixtures are fictional; no live accounts or browser actions are needed.

Cases 01–06 closely follow worked examples and serve as regression checks, not held-out evidence. Do not claim they demonstrate generalization or a with-plugin advantage. Cases 07–11 add different failure modes. The separate repository-level `evals/` cases also informed plugin development and are development and regression checks, not independent evidence of improvement. For comparisons, report the exact source revision, model settings, runs, and baseline. No model evaluation has been run merely because these files exist.

| Case | Focus | Raw packet |
| --- | --- | --- |
| 01 | Product concepts and alternate role needs | [Proposal structure](fixtures/01-product-model.md) |
| 02 | Decision information and consequential defaults | [Plan choice](fixtures/02-decision.md) |
| 03 | Visible routes versus source knowledge | [Find an invoice](fixtures/03-findability.md) |
| 04 | Labels, state, persistence, and consequences | [Keep a draft](fixtures/04-language.md) |
| 05 | Ownership and cross-role completion | [Request handoff](fixtures/05-handoff.md) |
| 06 | First-product-use and repeat efficiency | [Dispatch schedule](fixtures/06-progression.md) |
| 07 | Misleading success and actual task outcome | [Failed save](fixtures/07-false-success.md) |
| 08 | Counterexample to visual simplification | [Clear dense comparison](fixtures/08-density-counterexample.md) |
| 09 | Shared data versus resumable task context across devices | [Cross-device continuation](fixtures/09-cross-device-continuation.md) |
| 10 | Product promise versus first cross-platform task | [Promise-to-product continuity](fixtures/10-promise-to-product.md) |
| 11 | Counterexample to unnecessary synchronization | [Local-only requirements](fixtures/11-local-only-counterexample.md) |

For assessment after capture, use [the assessor rubric](assessor-rubric.md). For controlled comparisons, use [the record template](comparison-record.md).