Skip to content

The ultimate harness for coding agents and beyond. All your agents. All your machines. One command center. Start with code, then follow your curiosity and build across disciplines: CAD, circuits, robots, games and music.

Verify recovery with updated agent machine defaults

dee3a4ae2200298
diff --git a/desktop/test/agent_events_test.dart b/desktop/test/agent_events_test.dart
index 028179a59..29ee685ab 100644
--- a/desktop/test/agent_events_test.dart
+++ b/desktop/test/agent_events_test.dart
@@ -168,7 +168,9 @@ void main() {
await tester.pumpAndSettle();
await tester.tap(find.widgetWithText(FilledButton, 'Select this folder'));
await tester.pumpAndSettle();
- await tester.tap(find.widgetWithText(FilledButton, 'Create agent'));
+ await tester.tap(
+ find.widgetWithText(FilledButton, 'Create on ${machine.displayName}'),
+ );
await tester.pumpAndSettle();
}
diff --git a/desktop/test/new_agent_engine_check_test.dart b/desktop/test/new_agent_engine_check_test.dart
index 481df5c4e..01408a6f8 100644
--- a/desktop/test/new_agent_engine_check_test.dart
+++ b/desktop/test/new_agent_engine_check_test.dart
@@ -216,7 +216,7 @@ void main() {
await tester.pump();
await tester.tap(find.byKey(const Key('new-agent-machine-field')));
await tester.pumpAndSettle();
- await tester.tap(find.text('Other computer'));
+ await tester.tap(find.text('Other computer — Remote'));
await tester.pumpAndSettle();
await tester.tap(find.byKey(const ValueKey('new-agent-quick-codex')));
await tester.tap(find.byKey(const Key('new-agent-advanced')));
diff --git a/desktop/test/swarm_multi_add_test.dart b/desktop/test/swarm_multi_add_test.dart
index a18164d63..3e896ea5d 100644
--- a/desktop/test/swarm_multi_add_test.dart
+++ b/desktop/test/swarm_multi_add_test.dart
@@ -31,7 +31,7 @@ void main() {
expect(app.panes, [pane]);
await chord(tester, LogicalKeyboardKey.keyN, shift: true);
expect(find.byType(AlertDialog), findsOneWidget);
- expect(find.text('Create agent'), findsOneWidget);
+ expect(find.text('Create on Test host'), findsOneWidget);
expect(app.panes, [pane]);
expect(input, isEmpty);
await tester.tap(find.text('Cancel'));
diff --git a/docs/harness-v2-handoff.md b/docs/harness-v2-handoff.md
index 2d5b053c6..b0f274fbf 100644
--- a/docs/harness-v2-handoff.md
+++ b/docs/harness-v2-handoff.md
@@ -48,7 +48,9 @@ Research must lead to justified improvements, not feature accumulation.
the earlier preference for creating a fresh feature branch after each merge.
- **`d727360`** clarifies view removal and restores closed work into its existing
swarm. **`332546f`** keeps terminal relayout from taking keyboard focus from
- resize controls. The combined desktop suite passes 1,317 tests, one skip.
+ resize controls. **`eef17f6`** preserves the team's `e68c893` local-machine
+ default for general New agent actions; explicit split context stays intact.
+ The combined desktop suite passes 1,317 tests, one skip.
- **`339f008`** saves the opening continuation: it keeps the retained canvas
built while a picker opens/closes and shares validated catalogs across openings. Warm Add
opening measured 22.006 ms median versus 26.273 ms in the same headless
@@ -287,12 +289,13 @@ unverified; do not infer them from these UI changes.
**Archived drafts:** both are superseded. The onboarding ideas were adapted into
the common page, without adding another first-tab-only component.
-**Latest verification:** the History recovery and viewport-focus continuation on
-`e909396` passes **1,317 desktop tests**, one existing skip. The analyzer reports
+**Latest verification:** the History recovery and viewport-focus continuation,
+including the team's default-machine update in `eef17f6`, passes **1,317 desktop
+tests**, one existing skip. The analyzer reports
zero errors/warnings and the same 14 informational diagnostics. The isolated
AppKit titlebar/menu fixture passes 335 checks, including hidden window layout.
-Logs: `/private/tmp/harness-reopen-existing-verified-full-tests.log`,
-`/private/tmp/harness-reopen-existing-analyze.log` and
+Logs: `/private/tmp/harness-reopen-existing-final-full-tests.log`,
+`/private/tmp/harness-reopen-existing-final-analyze.log` and
`/private/tmp/harness-reopen-existing-native.log`. The 54 focused History checks
and 23 focused terminal/recovery checks also pass. No real agent received test
input; the running preview was not restarted. These changes have not had a new
diff --git a/docs/harness-v2-progress.md b/docs/harness-v2-progress.md
index 905ff47b8..fbb0233c7 100644
--- a/docs/harness-v2-progress.md
+++ b/docs/harness-v2-progress.md
@@ -20,8 +20,10 @@ Updated 2026-09-14 after integrating the continuation into main. This is a worki
switches to a valid four-pane layout. Archive/Resume is still a proposal;
current user runtimes were not started, stopped or used as test fixtures.
Verification: **1,317 desktop tests pass**, one existing skip; zero analyzer
- errors/warnings, 14 existing infos; 335 isolated AppKit checks pass. Logs:
- `/private/tmp/harness-reopen-existing-{verified-full-tests,analyze,native}.log`.
+ errors/warnings, 14 existing infos; 335 isolated AppKit checks pass. The final
+ combined suite includes the team's `e68c893` local-machine default, preserved
+ in merge `eef17f6`; existing fixtures now use its explicit remote labels.
+ Logs: `/private/tmp/harness-reopen-existing-{final-full-tests,final-analyze,native}.log`.
The running preview was not restarted; this continuation has no new Release
build or live native workflow result. Benchmarking remains deferred.
- Added **Retry** to failed New agent availability checks. The same form keeps