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.

Record phone PR status and verified build 48 installation

deehwebf660a94d0f
diff --git a/docs/research/2026-09-28-phone-overnight.md b/docs/research/2026-09-28-phone-overnight.md
index f28fc3e01..c67e8f82d 100644
--- a/docs/research/2026-09-28-phone-overnight.md
+++ b/docs/research/2026-09-28-phone-overnight.md
@@ -166,9 +166,9 @@ These are AI persona reviews of offline artifacts. They establish the visual sco
### State
-- Local iPhone build **48** implements the user's desktop-parity request: separate Branch/Worktree controls and a Model chooser, all under collapsed Options. **1,636 tests** pass (28 conditional render skips), **28 screen renders** pass, and refreshed coverage is **88.9%** (18,622/20,957 lines, excluding `third_party/`). Analysis has only the same 12 third-party informational findings. [Brief team handoff](2026-09-28-mobile-team-handoff.md).
-- The user's build-46 review and a [twelve-hour change audit](2026-09-28-mobile-ui-change-audit.md) record the four corrections above, implemented for local iPhone build **47**. They supersede conflicting panel recommendations. Follow-up validation: **1,621 tests** and **25 renders** pass; analyzer clean outside the same 12 third-party informational findings. The coverage figures above were measured on build 46. The user requested an iPhone install only, not TestFlight.
-- Branch `phone-overnight-polish`; [PR #398](https://github.com/autonomous-ai/openharness/pull/398) contains the continuation and main integration. The user explicitly authorized merging and a local iPhone install for manual review. Check the PR for the resulting merge commit; no store release was requested.
+- [PR #402](https://github.com/autonomous-ai/openharness/pull/402) implements the user's desktop-parity request: separate Branch/Worktree controls and a Model chooser, all under collapsed Options. **1,636 tests** pass (28 conditional render skips), **28 screen renders** pass, and refreshed coverage is **88.9%** (18,622/20,957 lines, excluding `third_party/`). Analysis has only the same 12 third-party informational findings. Signed build **48** is installed and its version verified on the review iPhone. [Brief team handoff](2026-09-28-mobile-team-handoff.md).
+- The user's build-46 review and a [twelve-hour change audit](2026-09-28-mobile-ui-change-audit.md) record the four corrections above, merged in [PR #401](https://github.com/autonomous-ai/openharness/pull/401) for local iPhone build **47**. They supersede conflicting panel recommendations. Follow-up validation: **1,621 tests** and **25 renders** pass; analyzer clean outside the same 12 third-party informational findings. The original overnight coverage figures were measured on build 46. The user requested an iPhone install only, not TestFlight.
+- [PR #398](https://github.com/autonomous-ai/openharness/pull/398) merged the overnight polish, cleanup and main integration. The user explicitly authorized the merges and local iPhone installs for manual review; no store release was requested.
- Merged into it and finished: `coverage-rest` (coverage engineer), `desktop-cut` (the desktop's half of the notifier), and the phone-screens and state-core engineers' passes. No engineer is still running.
- Never commit `mobile/ios/Runner.xcodeproj/project.pbxproj`. It carries the local signing team and stays modified in the worktree.
@@ -193,7 +193,7 @@ The full tests run from `mobile/` in the monorepo: protocol checks read CLI sour
From `mobile/`, with Flutter 3.47.2:
-- **Everything:** `flutter test`, about one minute. For coverage, `flutter test --coverage`, then read `coverage/lcov.info`.
+- **Everything:** `flutter test`, about two minutes. For coverage, `flutter test --coverage`, then read `coverage/lcov.info`.
- **Screen renders:** `PHONE_RENDER_DIR=<dir> flutter test test/render/phone_screens_render_test.dart`. It writes 28 PNGs, including expanded Options and Model selection. The fixture terminal is 42 columns, so canned lines must fit in 42.
- **Walk-through on the iOS simulator, sample mode only:** `HARNESS_JOURNEY_OUT=<dir> flutter drive --driver=test_driver/journey_driver.dart --target=integration_test/tour_test.dart`.
@@ -239,5 +239,5 @@ From `mobile/`, with Flutter 3.47.2:
- **Desktop [PR #393](https://github.com/autonomous-ai/openharness/pull/393)** is merged: Add Phone lists/removes paired devices; password pairing retains the authenticated device name. Validated on current main with 18 desktop tests, 23 CLI pairing tests and clean analysis of the changed desktop files.
- **Mobile PR #321** (clipboard image paste) remains open. The user limited merges to their own PRs; this PR belongs to another author and was not changed.
-- **Website `/pair` page:** PR autonomous-ai/autonomous-code#4 needs a `_web` tag to deploy.
-- **TestFlight** is blocked until someone is on the release Mac, which has the App Store Connect key. Until then, the phone is a local `devicectl` install only.
+- **Website [PR #4](https://github.com/autonomous-ai/autonomous-code/pull/4)** is merged: the `/pair` landing page and phone setup redirects. Its page test, redirect assertions and production build pass; changed files have no lint errors. Deployment still needs a `_web` release tag; no website release was made.
+- **TestFlight** still needs production signing and App Store Connect credentials on the release machine. The user requested local iPhone installation only; no TestFlight upload was made.