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.
TypeScript 3d-modelingagent-frameworkagent-harnessai-agentscadclaude-codecodexcoding-agentsdeveloper-toolsesp32game-developmentmcp-servermulti-agentopen-hardwareopencodeopenharnesspcbroboticsterminaltmux
Record current native fixture and picker checkpoint
dee730e2b362170
diff --git a/docs/harness-v2-handoff.md b/docs/harness-v2-handoff.mdindex a7d33dded..2fc9d3215 100644--- a/docs/harness-v2-handoff.md+++ b/docs/harness-v2-handoff.md@@ -30,7 +30,7 @@ Research must lead to justified improvements, not feature accumulation.- Continue on **`main`**, tracking `origin/main`. The user's latest instructionis to work, commit, and push directly on main from now on. This supersedesthe earlier preference for creating a fresh feature branch after each merge.-- The latest opening continuation keeps the retained canvas built while a+- **`339f008`** saves the opening continuation: it keeps the retained canvas built while apicker opens/closes and shares validated catalogs across openings. Warm Addopening measured 22.006 ms median versus 26.273 ms in the same headlessfixture; cold starts and native display latency are not established by it.@@ -299,11 +299,14 @@ a distinct Harness Benchmark app. Nine isolated Python checks pass. Preflightdistinguishes installed copies from workspace builds by exact location as wellas identity. Preserve its isolation and foreground/key-window guards. Theperformance notes explain the earlier failed calibration; these tooling fixes-provide no new latency measurements. The identity-fix disposable Release build-succeeded at `/private/tmp/harness-native-benchmark-7hv6voyr`, before the newer-Add frame/focus changes. Rebuild it for current-source timing. Its runner correctly-refused to start while the workspace preview was running, writing no timing-result. The [performance record](harness-v2-performance.md) has the artifacts.+provide no new latency measurements. The current disposable Release build+succeeded at `/private/tmp/harness-native-benchmark-wt0_dt31`, with production+source at `339f008`. The four changed picker source files match by SHA-256;+the built bundle is `ai.autonomous.harness.benchmark`. Build receipt:+`/private/tmp/harness-native-picker-prepare.log`. The earlier identity-fix runner+correctly refused the running workspace preview, writing no timing result.+This session confirmed that preview is still running; the new fixture was not+launched. The [performance record](harness-v2-performance.md) has the artifacts.## Previous terminal fix and verification@@ -383,10 +386,10 @@ that its isolated window can become active/key.using the repaired benchmark identity checks for the team's new`ai.autonomous.harness` macOS bundle identifier (Linux now uses`com.autonomous.harness` and executable `harness`). Nine isolation tests pass;- an identity-fix disposable Release fixture is available at- `/private/tmp/harness-native-benchmark-7hv6voyr`. Its runner correctly refuses- the running workspace preview. Rebuild it for the newer Add frame/focus- source. A request to briefly close/reopen that preview+ a current disposable Release fixture is available at+ `/private/tmp/harness-native-benchmark-wt0_dt31` from source `339f008`.+ Rebuild it if production source changes. The workspace preview is still+ running. A request to briefly close/reopen that previewfor calibration is pending; elapsed time is not approval. No native sampleshave been accepted. See the current performance record for exact artifacts.Then close remaining visual/platform/release qualification gaps. Maintaindiff --git a/docs/harness-v2-performance.md b/docs/harness-v2-performance.mdindex 56961a61e..f388f2e50 100644--- a/docs/harness-v2-performance.md+++ b/docs/harness-v2-performance.md@@ -117,9 +117,17 @@ correctly refused the still-running workspace preview at its exact build path;the installed copy is classified separately. No native samples were taken.The session requested a brief preview-close/idle window for calibration andcontinues independent performance work while that request is pending.-Rebuild the fixture before measuring the newer Add frame/focus changes above.Artifacts: `/private/tmp/harness-native-identity-{before,tests,prepare,preflight}.log`.+After the picker-opening continuation, a fresh isolated arm64 Release fixture+built from production source at `339f008`:+`/private/tmp/harness-native-benchmark-wt0_dt31/desktop/build/macos/Build/Products/Release/Harness Benchmark.app`.+The four changed picker files match the checkout by SHA-256; the built app has+the distinct `ai.autonomous.harness.benchmark` identity. Build receipt:+`/private/tmp/harness-native-picker-prepare.log`. The workspace preview is still+running and the close/reopen request is pending. This fixture was not launched;+no native samples were collected. Rebuild it if production source changes.+### History focus hot pathHistory now formats only agents with open views. Previously each focus changediff --git a/docs/harness-v2-progress.md b/docs/harness-v2-progress.mdindex 9d7141101..7511d9541 100644--- a/docs/harness-v2-progress.md+++ b/docs/harness-v2-progress.md@@ -17,6 +17,11 @@ Updated 2026-09-14 after integrating the continuation into main. This is a workiteam's independent device UI change `5f8baac` by fast-forwarding main.The normal macOS arm64 Release build succeeds. Verification artifacts:`/private/tmp/harness-add-open-{full-tests,analyze,release-build}.log`.+ Saved and pushed as `339f008`. A fresh isolated native Release fixture also+ built at `/private/tmp/harness-native-benchmark-wt0_dt31`; its picker files+ match this source and its bundle has the distinct benchmark identity. It was+ not launched: the preview is still running and the brief close/reopen request+ remains pending. Receipt: `/private/tmp/harness-native-picker-prepare.log`.- Improved Add's arrow-key frame work by reusing unchanged result rows andleaving the text editor alone when only the highlight changes. In the paired2,000-agent headless fixture, arrow selection median fell 8.592 to 4.714 ms;