Evidence-backed native dashboard proposal · The Builder is drafting a code proposal.

Set the first milestone as a proposed native Lovelace dashboard example, not a custom card. This follows the Builder’s narrower approach while accepting the Reviewer’s objection: documented entities alone do not establish reliable freshness or authentication states. Begin by inspecting coordinator, sensor, service, and agent-tool paths. The commit summary names an order-placement service, so excluding checkout controls from the dashboard must not be presented as integration-wide enforcement. This contribution will offer no ordering actions; cart review and checkout remain in Swiggy. Proceed with a small state contract, dashboard example, setup guidance, and proposed fixture tests. If required freshness or authentication signals are absent, document the gap and propose the smallest prerequisite separately rather than displaying unsupported assurances. MIT licensing is documented. Specific maintainer agreement is not established, and push access is unavailable. This is a scoped contribution proposal, not an approved release.

I support the Builder’s native Lovelace example, but challenge the assumption that existing entities alone can support a trustworthy dashboard. The supplied file listing does not establish whether failed refreshes preserve old values or expose a last-successful-update timestamp. A dashboard could therefore look healthy while showing an outdated total or ETA. Before choosing presentation, inspect the coordinator and sensor implementations for availability, authentication failures, partial refreshes, and timestamp semantics. Home Assistant entity-change time should not be treated as proof of a successful refresh: unchanged data can still be fresh. Proposed fixtures should distinguish an empty cart, no active delivery, missing ETA, expired authentication, and stale cached data. The Scout’s checkout concern remains important: the commit summary names an order-placement service, while the mission requires checkout in Swiggy. Omitting that control from the dashboard is appropriate, but does not establish that other service or agent-tool paths respect the same boundary; inspect those paths before describing the experience as review-only. The README’s session-conflict warning also makes checkout handoff and subsequent recovery an evidence gap, not merely a documentation detail. MIT licensing is supplied; specific maintainer agreement for this direction is not established.

The unfinished dashboard is a useful small-release opportunity, but I would start with a native Lovelace configuration example rather than a custom card. The README already documents cart, total, order status, and ETA entities; reusing those could keep the first contribution narrow and easier to review. The commit’s addition of `place_instamart_order` conflicts with the mission’s checkout boundary. This dashboard should expose no ordering action: users review and complete checkout in Swiggy. My next contribution would be a proposed dashboard example paired with a short state contract: empty cart is distinct from unavailable data; expired sessions prompt reauthentication; stale values are visibly marked with last-successful-update information where supported. Missing ETA must not become “0 minutes.” Whether the coordinator exposes enough information remains unverified from the supplied file listing. A useful review check would be fixture-based cases for those states, plus setup and Swiggy-handoff instructions addressing the README’s session-conflict warning. These are proposed checks, not executed tests. MIT licensing is documented. Specific maintainer agreement and permission to push are not established; this remains a contribution proposal.

The clearest unfinished opportunity is a trustworthy Instamart dashboard: the README lists cart, total, status, and ETA entities but still marks the Lovelace card unfinished. One concrete mismatch deserves attention before designing controls: the latest commit describes adding `place_instamart_order`, while the mission requires review and checkout in Swiggy. I would explore a read-only dashboard with an explicit Swiggy checkout handoff, rather than exposing order placement. My Scout concern is freshness, not just presentation. The README warns that simultaneous Swiggy app use can cause session conflicts—yet opening Swiggy is necessary for checkout. That transition needs investigation so an old cart total or ETA cannot appear current after a session failure. A useful next inquiry is whether existing coordinator data distinguishes an empty cart from unavailable data, expired authentication, and a stale successful response. Those distinctions could guide dashboard states, recovery instructions, and proposed fixture-based tests without assuming backend changes are necessary. MIT licensing is supplied; specific maintainer agreement for this dashboard direction is not established.