# Consumer journeys

Start with the person's task, then pin the SDK and choose the renderer. Resolve the pattern, name the product facts behind each visible state, compose with the supported entries and exercise the interaction on the declared target. Send shared defects back with a minimal SDK reproduction. [Design system rules](../../../design-system-rules.md) · [Evaluation protocol](evaluation.md)

Use independent synthetic consumers with invented records, local state and the pinned SDK. These journeys exercise transferable UI contracts. Additional projects can qualify the same patterns with their own facts and targets.

| Step | Review a revisioned decision | Inspect a partially observed record |
| --- | --- | --- |
| Task | Review a brief and decide whether to return it for revision. Find the original result after a disconnect. | Inspect a board with partial readings and understand what evidence is available. |
| Find the design | Use `pattern.review-decision` and `pattern.track-request`. Resolve the installed decision review, receipt and progress entries. | Use `pattern.inspect-record`. Resolve the installed chart, readable values and record facts. |
| Keep product ownership explicit | The host supplies the brief, revision, scope, permitted actions and original receipt lookup. The shared interaction conveys confirmation, pending and final result. | The host supplies timestamps, observation values, missing observations and current access. The shared presentation conveys which facts are observed or unavailable. |
| Exercise the task | Cancel and verify zero writes. Confirm the selected target and revision once. Lose the response, reopen and resolve the same receipt. Inspect the visible and accessible status before and after confirmation. | Read an observed zero and a missing value in the chart and text. Withdraw access while the record is open. Verify the projection and accessible values disappear. |
| Route a change | A wrong revision or receipt belongs to the host project. Reproduce shared focus loss on the pinned dialog before proposing a Yogurt fix. | A wrong observation or stale entitlement belongs to the host project. Reproduce skipped chart gaps in the shared entry before opening a Yogurt report. |

Run the decision fixture on web and on the Native host when evaluating support for both targets. Run the record fixture on web and record the matching Native support status. Exercise each scene from an isolated installation and attach the source checks, recorded actions and runtime observations to the exact SDK revision.

The Native record fixture uses the installed bar-chart and data-table templates. The pinned line renderer connects finite points across a missing sample. The bar chart keeps the original sample slots, while the table distinguishes observed zero from an unavailable reading. Keep the source and original time attached to the projection. Use a fixture-only access withdrawal event. On withdrawal the model clears the projection and resets disclosure. The rendered chart, table, source and values disappear from the Native accessibility snapshot. Attempting to reopen the table before the host restores access has no effect.

Run on the supported macOS Native SDK and Zig pins with two observations and one missing reading. Assert the state transitions in the model, then open the table and withdraw access in a real window. Capture the accessible snapshot and frame before and after withdrawal. Record chart and table support separately from any unexercised device or OS.
