You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Apply the selected appearance to the compatibility workspace #280
Part of #120. This delivery does not retire an adoption or release gate.
Evidence and audience
The kit's visible Dark selection changes the embedded product hint while its own workspace stays forcibly light. This leaves the author workflow inconsistent with the requested dark mode and the host-theme control. This affects product authors and consumers using the documented workflow.
Intended outcome
Make the declared workflow reliable without widening source access or moving data or credentials into controller metadata. A focused failing regression is the acceptance proxy; implementation tests do not establish independent production adoption.
Acceptance criteria
Apply light and dark appearance to the workspace using the established orange palette.
Keep host appearance independent of optional product hints and their grant intersection.
Verify keyboard controls, denied storage, narrow layouts and both product-appearance gate states.
Delivered by merged PR #282 at source 19c7ec790522f4a6ddf819929de86ee3953a951a, with reviewed and tested head 1d5621a06e0c01e8f85d3de00627ba4e57661764.
All 38 hosted results settled: 36 passed and two expected automation skips. CI exercised controller/API/chart checks, TLS browser interaction, real SQL/JSONB/AGE, Document and Graph queries and Kubernetes recovery lifecycle. The five deterministic cancellation-boundary regressions were RED before the repair and GREEN under 100 race iterations, while real connector timeout failures remain observable.
The complete local build/race/browser/provider-fixture/lint and chart contract validation passed. Exact-head review is recorded as GitHub review 5409858648 after verified provider usage limits, with an independent clean review of all eight paths.
This completes the source delivery. Production upgrade and authenticated public user-path verification remain tracked separately in devantler-tech/platform#4499; production adoption gates remain unchanged.
Activity
chore: agent-claim #280 nonce=c86894f32a16716bfc5c056eeb62a426
devantler commentedon Oct 5, 2026
Delivered by merged PR #282 at source
19c7ec790522f4a6ddf819929de86ee3953a951a, with reviewed and tested head1d5621a06e0c01e8f85d3de00627ba4e57661764.All 38 hosted results settled: 36 passed and two expected automation skips. CI exercised controller/API/chart checks, TLS browser interaction, real SQL/JSONB/AGE, Document and Graph queries and Kubernetes recovery lifecycle. The five deterministic cancellation-boundary regressions were RED before the repair and GREEN under 100 race iterations, while real connector timeout failures remain observable.
The complete local build/race/browser/provider-fixture/lint and chart contract validation passed. Exact-head review is recorded as GitHub review 5409858648 after verified provider usage limits, with an independent clean review of all eight paths.
This completes the source delivery. Production upgrade and authenticated public user-path verification remain tracked separately in devantler-tech/platform#4499; production adoption gates remain unchanged.