Repository navigation
fix(agent): reduce suggestions in narrow panels - #18384
Conversation
📦 Bundle: 9.49 MB gzip 🔴 +1 BDetailsSummary
Category Glance App Entry Points — 6.65 kB (baseline 6.65 kB) • ⚪ 0 BMain entry bundles and manifests
Status: 1 added / 1 removed Graph Workspace — 3.47 MB (baseline 3.47 MB) • ⚪ 0 BGraph editor runtime, canvas, workflow orchestration Status: 4 unchanged Views & Navigation — 147 kB (baseline 147 kB) • ⚪ 0 BTop-level views, pages, and routed surfaces Status: 18 unchanged Panels & Settings — 674 kB (baseline 674 kB) • ⚪ 0 BConfiguration panels, inspectors, and settings screens Status: 30 unchanged User & Accounts — 151 kB (baseline 151 kB) • ⚪ 0 BAuthentication, profile, and account management bundles Status: 17 unchanged Editors & Dialogs — 359 kB (baseline 359 kB) • ⚪ 0 BModals, dialogs, drawers, and in-app editors Status: 12 unchanged UI Components — 212 kB (baseline 212 kB) • ⚪ 0 BReusable component library chunks Status: 24 unchanged Data & Services — 1.11 MB (baseline 1.11 MB) • ⚪ 0 BStores, services, APIs, and repositories Status: 37 unchanged Utilities & Hooks — 993 kB (baseline 993 kB) • ⚪ 0 BHelpers, composables, and utility bundles Status: 72 unchanged Vendor & Third-Party — 13.9 MB (baseline 13.9 MB) • ⚪ 0 BExternal libraries and shared vendor chunks Status: 17 unchanged Other — 19.7 MB (baseline 19.7 MB) • ⚪ 0 BBundles that do not match a named category
Status: 1 added / 1 removed / 366 unchanged ⚡ Performance Report
Scenario timings
Show regressions
All metrics
Historical variance (last 15 runs)
Trend (last 15 commits on main)
Raw data{
"timestamp": "2026-10-17T10:46:27.202Z",
"gitSha": "d3589cc92f95f90915ec54083ef7555d87aa94ec",
"branch": "christian-byrne/pm-1373-compact-suggestions",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2007.4280000000044,
"styleRecalcs": 10,
"styleRecalcDurationMs": 3.2970000000000006,
"layouts": 10,
"layoutDurationMs": 2.979,
"taskDurationMs": 586.4079999999999,
"heapDeltaBytes": -465164,
"heapUsedBytes": 63661536,
"domNodes": -162,
"jsHeapTotalBytes": 7340032,
"scriptDurationMs": 11.398000000000001,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666942148760334,
"p95FrameDurationMs": 16.799999999999727
},
{
"name": "canvas-idle",
"durationMs": 2040.616,
"styleRecalcs": 10,
"styleRecalcDurationMs": 3.0279999999999996,
"layouts": 9,
"layoutDurationMs": 3.068,
"taskDurationMs": 618.599,
"heapDeltaBytes": -4836776,
"heapUsedBytes": 59653508,
"domNodes": -162,
"jsHeapTotalBytes": 7340032,
"scriptDurationMs": 13.392999999999999,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66639344262295,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1840.0490000000218,
"styleRecalcs": 77,
"styleRecalcDurationMs": 14.388000000000002,
"layouts": 10,
"layoutDurationMs": 3.011,
"taskDurationMs": 946.9609999999999,
"heapDeltaBytes": 15011844,
"heapUsedBytes": 79472120,
"domNodes": -162,
"jsHeapTotalBytes": 7864320,
"scriptDurationMs": 168.854,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666363636363638,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-mouse-sweep",
"durationMs": 2043.7479999999368,
"styleRecalcs": 84,
"styleRecalcDurationMs": 15.996999999999998,
"layouts": 9,
"layoutDurationMs": 2.4270000000000005,
"taskDurationMs": 1143.087,
"heapDeltaBytes": 16026820,
"heapUsedBytes": 80508952,
"domNodes": -162,
"jsHeapTotalBytes": 7077888,
"scriptDurationMs": 173.297,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.6655737704918,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1725.9189999999762,
"styleRecalcs": 18,
"styleRecalcDurationMs": 6.034,
"layouts": 11,
"layoutDurationMs": 3.5330000000000004,
"taskDurationMs": 333.1,
"heapDeltaBytes": 24466480,
"heapUsedBytes": 88906376,
"domNodes": 47,
"jsHeapTotalBytes": 7340032,
"scriptDurationMs": 15.044000000000002,
"eventListeners": 37,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666346153846153,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1707.5439999999844,
"styleRecalcs": 19,
"styleRecalcDurationMs": 5.561,
"layouts": 10,
"layoutDurationMs": 2.851,
"taskDurationMs": 465.57400000000007,
"heapDeltaBytes": 24035868,
"heapUsedBytes": 88523716,
"domNodes": 47,
"jsHeapTotalBytes": 7077888,
"scriptDurationMs": 15.284000000000002,
"eventListeners": 37,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.6656862745098,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 707.9939999999851,
"styleRecalcs": 13,
"styleRecalcDurationMs": 2.974,
"layouts": 9,
"layoutDurationMs": 2.418,
"taskDurationMs": 516.612,
"heapDeltaBytes": 7050704,
"heapUsedBytes": 85841608,
"domNodes": 0,
"jsHeapTotalBytes": 8126464,
"scriptDurationMs": 175.08499999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 776.2189999999691,
"styleRecalcs": 13,
"styleRecalcDurationMs": 3.0960000000000005,
"layouts": 8,
"layoutDurationMs": 2.806,
"taskDurationMs": 542.771,
"heapDeltaBytes": 298768,
"heapUsedBytes": 63398364,
"domNodes": 0,
"jsHeapTotalBytes": 3407872,
"scriptDurationMs": 176.94899999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.667391304347824,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1147.5269999999682,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 825.7139999999999,
"heapDeltaBytes": 17214580,
"heapUsedBytes": 96303584,
"domNodes": 0,
"jsHeapTotalBytes": 17989632,
"scriptDurationMs": 64.49399999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 512,
"frameDurationMs": 33.82058823529411,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-dense-n200-w200-i120",
"durationMs": 1157.2219999999334,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 837.4370000000001,
"heapDeltaBytes": 16795388,
"heapUsedBytes": 97113096,
"domNodes": 0,
"jsHeapTotalBytes": 16941056,
"scriptDurationMs": 67.732,
"eventListeners": 0,
"totalBlockingTimeMs": 522,
"frameDurationMs": 33.8235294117647,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 858.9509999999905,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 521.4580000000004,
"heapDeltaBytes": -4543544,
"heapUsedBytes": 74630844,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 160.984,
"eventListeners": 0,
"totalBlockingTimeMs": 218,
"frameDurationMs": 24.76285714285711,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-order-sparse-n1000-w20-i120",
"durationMs": 844.8610000000372,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 500.1460000000004,
"heapDeltaBytes": -4105220,
"heapUsedBytes": 74627624,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 162.94200000000004,
"eventListeners": 0,
"totalBlockingTimeMs": 222,
"frameDurationMs": 24.9970588235294,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "extension-clean-combined-draw",
"durationMs": 11161.923000000003,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10188.703,
"heapDeltaBytes": 13750144,
"heapUsedBytes": 97870036,
"domNodes": 0,
"jsHeapTotalBytes": 5767168,
"scriptDurationMs": 8654.898000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 1752,
"frameDurationMs": 33.78666666666667,
"p95FrameDurationMs": 66.70000000000073
},
{
"name": "extension-clean-combined-draw",
"durationMs": 11296.19100000002,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10468.198,
"heapDeltaBytes": 5171488,
"heapUsedBytes": 88311852,
"domNodes": 0,
"jsHeapTotalBytes": 9961472,
"scriptDurationMs": 8569.398,
"eventListeners": 0,
"totalBlockingTimeMs": 1424,
"frameDurationMs": 31.962889518413597,
"p95FrameDurationMs": 66.69999999999891
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10223.969000000012,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 9380.805,
"heapDeltaBytes": 1146288,
"heapUsedBytes": 84314132,
"domNodes": 0,
"jsHeapTotalBytes": 3145728,
"scriptDurationMs": 7555.884000000001,
"eventListeners": 0,
"totalBlockingTimeMs": 771,
"frameDurationMs": 31.052583586626138,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "extension-rgthree-prototype-draw",
"durationMs": 10851.58100000001,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 10026.09,
"heapDeltaBytes": 11019952,
"heapUsedBytes": 94444100,
"domNodes": 0,
"jsHeapTotalBytes": 3145728,
"scriptDurationMs": 8147.516999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 1076,
"frameDurationMs": 31.26657060518732,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "large-combo-open",
"durationMs": 207.54099999999198,
"styleRecalcs": 23,
"styleRecalcDurationMs": 12.735,
"layouts": 10,
"layoutDurationMs": 2.915,
"taskDurationMs": 160.455,
"heapDeltaBytes": 13312200,
"heapUsedBytes": 78036004,
"domNodes": 917,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 9.113999999999999,
"eventListeners": 196,
"totalBlockingTimeMs": 0,
"frameDurationMs": 21.660000000000036,
"p95FrameDurationMs": 66.69999999999982
},
{
"name": "large-combo-open",
"durationMs": 218.7300000000505,
"styleRecalcs": 22,
"styleRecalcDurationMs": 13.607,
"layouts": 8,
"layoutDurationMs": 2.6120000000000005,
"taskDurationMs": 159.78599999999997,
"heapDeltaBytes": 13316808,
"heapUsedBytes": 77470800,
"domNodes": 917,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 8.998999999999999,
"eventListeners": 196,
"totalBlockingTimeMs": 1,
"frameDurationMs": 21.659999999999947,
"p95FrameDurationMs": 66.59999999999945
},
{
"name": "large-graph-idle",
"durationMs": 2014.8519999999621,
"styleRecalcs": 6,
"styleRecalcDurationMs": 1.5690000000000002,
"layouts": 5,
"layoutDurationMs": 3.388,
"taskDurationMs": 439.1590000000001,
"heapDeltaBytes": 10524304,
"heapUsedBytes": 74872512,
"domNodes": -258,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 33.303999999999995,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666115702479342,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-idle",
"durationMs": 2045.4280000000153,
"styleRecalcs": 6,
"styleRecalcDurationMs": 1.9789999999999999,
"layouts": 6,
"layoutDurationMs": 1.8179999999999998,
"taskDurationMs": 492.65899999999993,
"heapDeltaBytes": -25172572,
"heapUsedBytes": 74618976,
"domNodes": -278,
"jsHeapTotalBytes": -225280,
"scriptDurationMs": 35.794000000000004,
"eventListeners": -68,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666115702479335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-pan",
"durationMs": 2730.698000000018,
"styleRecalcs": 64,
"styleRecalcDurationMs": 5.023999999999999,
"layouts": 3,
"layoutDurationMs": 0.8439999999999993,
"taskDurationMs": 1693.3919999999998,
"heapDeltaBytes": 16369148,
"heapUsedBytes": 81606612,
"domNodes": 0,
"jsHeapTotalBytes": 225280,
"scriptDurationMs": 835.358,
"eventListeners": 2,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666463414634148,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2752.475000000004,
"styleRecalcs": 63,
"styleRecalcDurationMs": 5.338999999999999,
"layouts": 2,
"layoutDurationMs": 0.536,
"taskDurationMs": 1799.325,
"heapDeltaBytes": 15361248,
"heapUsedBytes": 81674008,
"domNodes": 0,
"jsHeapTotalBytes": 1011712,
"scriptDurationMs": 881.627,
"eventListeners": 2,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.870552147239266,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-zoom",
"durationMs": 3827.0899999999983,
"styleRecalcs": 67,
"styleRecalcDurationMs": 6.409,
"layouts": 61,
"layoutDurationMs": 9.996,
"taskDurationMs": 1775.488,
"heapDeltaBytes": 4109096,
"heapUsedBytes": 71109476,
"domNodes": 24,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 794.11,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665652173913045,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "large-graph-zoom",
"durationMs": 3614.9920000000293,
"styleRecalcs": 67,
"styleRecalcDurationMs": 7.347000000000001,
"layouts": 61,
"layoutDurationMs": 10.702,
"taskDurationMs": 1712.63,
"heapDeltaBytes": 3825400,
"heapUsedBytes": 71569384,
"domNodes": 24,
"jsHeapTotalBytes": 786432,
"scriptDurationMs": 831.021,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66574074074074,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "legacy-node-drag",
"durationMs": 2895.329000000004,
"styleRecalcs": 41,
"styleRecalcDurationMs": 3.913,
"layouts": 2,
"layoutDurationMs": 0.516,
"taskDurationMs": 1774.543,
"heapDeltaBytes": 7572884,
"heapUsedBytes": 74969648,
"domNodes": 0,
"jsHeapTotalBytes": 2621440,
"scriptDurationMs": 972.9870000000001,
"eventListeners": 182,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.95847953216374,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "legacy-node-drag",
"durationMs": 3137.890999999968,
"styleRecalcs": 39,
"styleRecalcDurationMs": 3.9539999999999993,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 2065.3379999999997,
"heapDeltaBytes": 18040796,
"heapUsedBytes": 86965136,
"domNodes": 0,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 989.2279999999998,
"eventListeners": 182,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.845161290322583,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2022.231999999974,
"styleRecalcs": 4,
"styleRecalcDurationMs": 0.9399999999999998,
"layouts": 3,
"layoutDurationMs": 0.778,
"taskDurationMs": 419.103,
"heapDeltaBytes": 10935412,
"heapUsedBytes": 76746452,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 37.20799999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665289256198346,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2010.9899999999925,
"styleRecalcs": 3,
"styleRecalcDurationMs": 0.6810000000000002,
"layouts": 2,
"layoutDurationMs": 0.6339999999999996,
"taskDurationMs": 672.4219999999999,
"heapDeltaBytes": 9732288,
"heapUsedBytes": 75149228,
"domNodes": 0,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 38.172,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665289256198346,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 695.0009999999907,
"styleRecalcs": 47,
"styleRecalcDurationMs": 5.488,
"layouts": 9,
"layoutDurationMs": 2.593,
"taskDurationMs": 495.914,
"heapDeltaBytes": 1684304,
"heapUsedBytes": 81986532,
"domNodes": 0,
"jsHeapTotalBytes": 8650752,
"scriptDurationMs": 151.721,
"eventListeners": 46,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.668292682926822,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 697.4199999999655,
"styleRecalcs": 46,
"styleRecalcDurationMs": 5.12,
"layouts": 9,
"layoutDurationMs": 2.995,
"taskDurationMs": 470.0249999999999,
"heapDeltaBytes": -1913548,
"heapUsedBytes": 63971148,
"domNodes": 0,
"jsHeapTotalBytes": 7602176,
"scriptDurationMs": 149.17999999999998,
"eventListeners": 46,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.668292682926822,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2023.2760000000098,
"styleRecalcs": 10,
"styleRecalcDurationMs": 3.2459999999999996,
"layouts": 10,
"layoutDurationMs": 3.054,
"taskDurationMs": 583.932,
"heapDeltaBytes": -19313612,
"heapUsedBytes": 61382572,
"domNodes": -180,
"jsHeapTotalBytes": 8126464,
"scriptDurationMs": 9.782,
"eventListeners": -114,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665289256198342,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2025.1739999999927,
"styleRecalcs": 11,
"styleRecalcDurationMs": 3.758,
"layouts": 10,
"layoutDurationMs": 3.599,
"taskDurationMs": 519.378,
"heapDeltaBytes": -16796512,
"heapUsedBytes": 63984844,
"domNodes": -180,
"jsHeapTotalBytes": 8126464,
"scriptDurationMs": 14.408999999999999,
"eventListeners": -114,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66611570247934,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 2008.7609999999927,
"styleRecalcs": 87,
"styleRecalcDurationMs": 16.872,
"layouts": 10,
"layoutDurationMs": 2.91,
"taskDurationMs": 1105.429,
"heapDeltaBytes": -2334120,
"heapUsedBytes": 78454068,
"domNodes": -180,
"jsHeapTotalBytes": 8650752,
"scriptDurationMs": 112.03200000000002,
"eventListeners": -114,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66583333333334,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 2025.9420000001,
"styleRecalcs": 89,
"styleRecalcDurationMs": 17.332,
"layouts": 10,
"layoutDurationMs": 2.887,
"taskDurationMs": 1204.512,
"heapDeltaBytes": 1361856,
"heapUsedBytes": 82021832,
"domNodes": -180,
"jsHeapTotalBytes": 8912896,
"scriptDurationMs": 122.31599999999999,
"eventListeners": -114,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.665289256198346,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "subgraph-transition-enter",
"durationMs": 1463.706000000002,
"styleRecalcs": 20,
"styleRecalcDurationMs": 27.486000000000004,
"layouts": 5,
"layoutDurationMs": 17.259999999999998,
"taskDurationMs": 1122.589,
"heapDeltaBytes": 2493256,
"heapUsedBytes": 99859580,
"domNodes": 15950,
"jsHeapTotalBytes": 16916480,
"scriptDurationMs": 31.058999999999997,
"eventListeners": 3244,
"totalBlockingTimeMs": 165,
"frameDurationMs": 19.55333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "viewport-pan-sweep",
"durationMs": 10599.092999999983,
"styleRecalcs": 246,
"styleRecalcDurationMs": 19.503,
"layouts": 4,
"layoutDurationMs": 1.065,
"taskDurationMs": 6311.9800000000005,
"heapDeltaBytes": -18632528,
"heapUsedBytes": 80977076,
"domNodes": -258,
"jsHeapTotalBytes": 491520,
"scriptDurationMs": 2917.9109999999996,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.692271293375395,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8718.232999999942,
"styleRecalcs": 246,
"styleRecalcDurationMs": 20.227999999999998,
"layouts": 3,
"layoutDurationMs": 0.8059999999999994,
"taskDurationMs": 5908.189,
"heapDeltaBytes": 17000480,
"heapUsedBytes": 81513556,
"domNodes": -258,
"jsHeapTotalBytes": 1241088,
"scriptDurationMs": 3083.128,
"eventListeners": -66,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.73019230769231,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 16549.60999999997,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 15420.821999999998,
"heapDeltaBytes": -3919700,
"heapUsedBytes": 284753268,
"domNodes": -3330,
"jsHeapTotalBytes": 22269952,
"scriptDurationMs": 256.77000000000004,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 28.261025641025643,
"p95FrameDurationMs": 50.100000000000364
},
{
"name": "vue-large-graph-idle",
"durationMs": 17133.28899999999,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16540.332,
"heapDeltaBytes": -19927596,
"heapUsedBytes": 269276516,
"domNodes": -3330,
"jsHeapTotalBytes": 20172800,
"scriptDurationMs": 326.14500000000004,
"eventListeners": -653,
"totalBlockingTimeMs": 0,
"frameDurationMs": 29.9520979020979,
"p95FrameDurationMs": 66.59999999999854
},
{
"name": "vue-large-graph-pan",
"durationMs": 19860.605999999963,
"styleRecalcs": 167,
"styleRecalcDurationMs": 14.679999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 19435.658,
"heapDeltaBytes": 6530108,
"heapUsedBytes": 279243552,
"domNodes": 0,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 838.441,
"eventListeners": 2,
"totalBlockingTimeMs": 35,
"frameDurationMs": 35.988949275362316,
"p95FrameDurationMs": 66.70000000000073
},
{
"name": "vue-large-graph-pan",
"durationMs": 28417.089000000033,
"styleRecalcs": 181,
"styleRecalcDurationMs": 18.957,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 28016.824,
"heapDeltaBytes": 6213140,
"heapUsedBytes": 267756920,
"domNodes": 0,
"jsHeapTotalBytes": 7077888,
"scriptDurationMs": 922.685,
"eventListeners": 2,
"totalBlockingTimeMs": 31,
"frameDurationMs": 52.29981583793739,
"p95FrameDurationMs": 100.10000000000218
},
{
"name": "vue-node-resize-workload",
"durationMs": 8470.578000000045,
"styleRecalcs": 20,
"styleRecalcDurationMs": 96.04800000000002,
"layouts": 20,
"layoutDurationMs": 23.281999999999996,
"taskDurationMs": 8321.542999999998,
"heapDeltaBytes": -28731096,
"heapUsedBytes": 269650556,
"domNodes": -3330,
"jsHeapTotalBytes": 14405632,
"scriptDurationMs": 163.44599999999997,
"eventListeners": -653,
"totalBlockingTimeMs": 111,
"frameDurationMs": 46.775138121546966,
"p95FrameDurationMs": 100
},
{
"name": "vue-node-resize-workload",
"durationMs": 8609.786999999982,
"styleRecalcs": 20,
"styleRecalcDurationMs": 97.26899999999999,
"layouts": 20,
"layoutDurationMs": 26.038999999999994,
"taskDurationMs": 8464.579,
"heapDeltaBytes": -59497240,
"heapUsedBytes": 242955308,
"domNodes": -3331,
"jsHeapTotalBytes": 4968448,
"scriptDurationMs": 189.13200000000003,
"eventListeners": -653,
"totalBlockingTimeMs": 123,
"frameDurationMs": 48.04245810055866,
"p95FrameDurationMs": 100.10000000000036
},
{
"name": "vue-zoom-out-idle",
"durationMs": 11685.928999999987,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 11506.503,
"heapDeltaBytes": -12710864,
"heapUsedBytes": 259698644,
"domNodes": 0,
"jsHeapTotalBytes": 786432,
"scriptDurationMs": 169.77900000000008,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 40.00993150684932,
"p95FrameDurationMs": 83.30000000000291
},
{
"name": "vue-zoom-out-idle",
"durationMs": 12163.50799999998,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 11791.228000000001,
"heapDeltaBytes": 14836256,
"heapUsedBytes": 263866512,
"domNodes": 0,
"jsHeapTotalBytes": 12582912,
"scriptDurationMs": 180.86599999999999,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 41.952068965517235,
"p95FrameDurationMs": 83.40000000000146
},
{
"name": "workflow-execution",
"durationMs": 124.72600000000966,
"styleRecalcs": 9,
"styleRecalcDurationMs": 10.711,
"layouts": 9,
"layoutDurationMs": 3.922,
"taskDurationMs": 95.307,
"heapDeltaBytes": 4796816,
"heapUsedBytes": 81214272,
"domNodes": 124,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 12.439999999999998,
"eventListeners": 1,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.662500000000023,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "workflow-execution",
"durationMs": 484.62100000006103,
"styleRecalcs": 12,
"styleRecalcDurationMs": 9.091000000000001,
"layouts": 11,
"layoutDurationMs": 3.8460000000000005,
"taskDurationMs": 130.11700000000002,
"heapDeltaBytes": -13382884,
"heapUsedBytes": 63090124,
"domNodes": 95,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 10.995000000000001,
"eventListeners": 1,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.662068965517236,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 462.2120000000223,
"styleRecalcs": 5,
"styleRecalcDurationMs": 0.773,
"layouts": 76,
"layoutDurationMs": 15.773,
"taskDurationMs": 406.75500000000005,
"heapDeltaBytes": 11025876,
"heapUsedBytes": 79504904,
"domNodes": 4,
"jsHeapTotalBytes": 786432,
"scriptDurationMs": 191.98399999999998,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.664285714285693,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "agent-composer-empty-conversation-typing",
"durationMs": 539.0310000000227,
"styleRecalcs": 5,
"styleRecalcDurationMs": 0.825,
"layouts": 76,
"layoutDurationMs": 16.892999999999997,
"taskDurationMs": 488.088,
"heapDeltaBytes": -14695392,
"heapUsedBytes": 78140880,
"domNodes": 4,
"jsHeapTotalBytes": 1048576,
"scriptDurationMs": 229.209,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.662499999999994,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 597.7420000000393,
"styleRecalcs": 5,
"styleRecalcDurationMs": 2.968,
"layouts": 76,
"layoutDurationMs": 15.91,
"taskDurationMs": 557.7710000000001,
"heapDeltaBytes": 341656,
"heapUsedBytes": 116822776,
"domNodes": 4,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 223.14100000000002,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "agent-composer-long-conversation-typing",
"durationMs": 623.4519999999293,
"styleRecalcs": 5,
"styleRecalcDurationMs": 2.299,
"layouts": 76,
"layoutDurationMs": 15.334999999999999,
"taskDurationMs": 585.833,
"heapDeltaBytes": 993896,
"heapUsedBytes": 117601692,
"domNodes": 4,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 239.958,
"eventListeners": 292,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.11351351351351,
"p95FrameDurationMs": 16.700000000000728
}
]
}🔬 E2E CoverageLine coverage: 64.8% Coverage totals
Top 10 files by uncovered lines
🎭 Playwright: ✅ 2282 passed, 0 failed · 2 flaky📊 Browser Reports
🎨 Storybook: ✅ Built — View Storybook🧩 Custom Nodes Ecosystem Matrix: 🟡 Withheld — could not measureExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight
Findings and check detailsThe matrix runs on same-repository version-bump PRs with relevant changes, main-branch pushes, and manual dispatches. This candidate does not qualify, so every expensive stage was skipped. Reproduce locally# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
npx vitest run --config vitest.matrix.config.mts
Full docs: scripts/registry-census/README.md 🌾 Fallow: ✅ PassedFallow audits the diff for duplication, complexity and dead code. It gates on new findings only, so when the audit completes, a red is something this PR introduced or moved — never inherited backlog. A red can also mean the audit itself did not finish; the details below say which happened. Findings and check detailsNo new findings in the changed files. Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since d965cd4dbd65b24546c518a0319f0f1edb3d4ada
Full docs: ADR-DEVEX-LINT-0015 |
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18384 +/- ##
==========================================
+ Coverage 86.58% 87.40% +0.82%
==========================================
Files 3750 2673 -1077
Lines 176944 148285 -28659
Branches 48825 40644 -8181
==========================================
- Hits 153201 129604 -23597
+ Misses 21676 18291 -3385
+ Partials 2067 390 -1677
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 1086 files with indirect coverage changes 🚀 New features to boost your workflow:
|
Playwright videos for new testsDownload the recorded videos and HTML report Recorded projects: cloud Recorded test titles:
|
Obsolete pre-merge check review: the latest head completed all required checks successfully and there are no actionable review threads.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: changes requested
Pre-merge blocker:
- [P1] The max query stops applying at 459px, so widths from 459px to just below 460px show all five suggestions even though the stated breakpoint is 460px. I reproduced this in Chromium and verified that Tailwind emits a strict
width < 459pxquery. The added tests do not exercise 459px or 460px, so they miss the defect.
Findings without an inline anchor
- [P3] The Testing section says the cloud Playwright job passed across all 16 shards. The head has one successful
playwright-tests (cloud, frontend-dist-cloud)check and 16 separately namedplaywright-tests-chromium-shardedchecks. Please name these as separate results so the test environment is accurate.
Existing feedback checked
- I contested the dismissed CodeRabbit review that says pre-merge checks failed. It applies to
e8eced7; the current head has 83 successful and 17 skipped check runs, with none failed or pending.
Unverified PR-description claim
- [P3] The supplied local evidence confirms oxfmt, ESLint, and stylelint, but not the statement that oxlint and knip ran through local hooks. The hooks are configured to run them, but a hook transcript is needed to establish that they ran locally. Please either provide that evidence or keep the local-testing bullet to the checks with recorded results.
Checked and cleared
- Fallow reported arbitrary-utility drift for the lightbulb icon, workflow icon, drop shadow, max width, and sparkles fallback. All five are present in the base file; the sparkles fallback only moved from line 64 to line 69. None was introduced by this PR.
Coverage
- Depth: standard.
- Lanes: correctness, maintainability, idiomaticity, tests, and documentation.
- Correctness lenses: blast radius, regression risk, idempotency, shared state, boundary discipline, and root-cause analysis.
- Maintainability lenses: ponytail ultra, no-comments, reader load, subtract-before-add, laziness, and legacy API migration.
- Test lenses: unit-test review, mock quality, proof, setup consolidation, flaky E2E review, Playwright-spec review, and Playwright authoring guidance.
- Checks: focused Vitest 2/2 passed; project and browser typechecks passed; changed-file ESLint, stylelint, and oxfmt passed; diff check passed; Fallow completed. The head has no failed or pending remote checks. A local cloud Playwright attempt did not reach the test because the dev server lacked the backend settings endpoint; the head's remote cloud run passed the test.
This comment has been minimized.
This comment has been minimized.
DrJKL
left a comment
There was a problem hiding this comment.
Verdict: follow-up fixes are correct; non-blocking cleanup remains
The breakpoint defect and all five inline findings from my prior review are fixed. The component now uses the complementary < 460px query, the browser case checks 459px and 460px, the class-token unit test is gone, and the page object owns panel opening and resizing against the actual queried root.
Findings without an inline anchor
- [P3] The Testing section still says the single cloud Playwright job passed across all 16 shards. The named SHA had one cloud check and 16 separately named
playwright-tests-chromium-shardedchecks. Please report those as separate configurations. - [P3] The browser case is still described as checking 420px and 480px, but the current test checks the exact 459px and 460px boundary.
- [P3] The focused Vitest result still says two tests passed. After deleting the class-token assertion, the current file has one test and the current-head run reports one passed.
- [P3] Both remote-result bullets still point to
860a4a9. Those checks predate the breakpoint, helper, and test fixes. Please update them to the current head after its checks complete, or label the current checks as in progress.
Prior findings
- Fixed: the 459px-to-460px breakpoint gap.
- Fixed: the class-token change-detector unit test.
- Fixed: loops in the Playwright test body.
- Fixed: bypassing
AgentPanel.open(). - Fixed: duplicated raw resize mechanics and the later outer-shell measurement error.
- Fixed: local oxlint/knip evidence. The author confirmed full pre-commit and pre-push checks, the hook configuration runs those checks, and this review's type-aware oxlint run passed.
- Remaining: the cloud/shard wording in the PR body.
Coverage
- Depth: standard follow-up.
- Lanes: correctness, maintainability, idiomaticity, tests, and documentation.
- Local checks at
95e0a47: full typecheck, changed-file ESLint, type-aware oxlint, full format check, focused Vitest 1/1, Playwright discovery of all six cloud tests, and diff check passed. - Regression proof: Tailwind compilation reproduced the old strict
< 459pxquery and the fixed strict< 460pxquery. Chromium rendered three suggestions at 459px and five at 460px. - Current remote checks at posting time: 71 successful, 17 skipped, six in progress, none failed.
- Browser execution was not run locally because the follow-up procedure requires CI evidence without starting a backend. The current-head Playwright checks remain in progress.
- Maintainability: all prior duplication and false-confidence findings are fixed; no new maintainability finding.
|
Follow-up complete on current head |
Dismissing as superseded by your own two follow-up reviews on this PR. Your 2026-09-22T15:33Z review recorded the 459px-to-460px breakpoint gap and all five inline findings as fixed, and your 2026-09-22T16:03Z review recorded the page-object and trace-phase findings as fixed with only a PR-body SHA nit left, which is now corrected. The branch has since been rebased from 7d166ce onto main at 1697904; new head f63a35c. Details and the new regression proof are in the PR comment.
|
@DrJKL this is the narrow-panel Agent suggestion fix from mid-September. It sat conflicting against The resolution is not a replay, because the file moved underneath it. All six cases in the spec file pass against a local backend here, and reverting only I dismissed your earlier changes-requested, quoting your own follow-ups that recorded the breakpoint defect and all five inline findings as fixed. No human had approved the PR, so the rebase cost no approval. |
There was a problem hiding this comment.
🔍 Cursor Review — Consolidated panel
Triggered by @christian-byrne.
Found 2 finding(s).
| Severity | Count |
|---|---|
| 🟡 Medium | 1 |
| 🟢 Low | 1 |
Panel: 6/6 reviewers contributed findings.
| :class=" | ||
| cn( | ||
| 'w-full max-w-full min-w-0 justify-start rounded-full px-3 text-sm @min-[460px]:w-auto', | ||
| index >= COMPACT_PROMPT_COUNT && '@max-[460px]:hidden' |
There was a problem hiding this comment.
🟡 Medium — Slots 3 and up are now hidden with CSS whenever the panel container is narrower than 460px. That is the default state: the dock defaults to 420px, which leaves about 402px of content. Telemetry is unaware of this. onPromptClick still reports promptCount: prompts.value.length (5), and the rendered experiment exposure still counts all 5 treatment prompts, so the starter-prompt A/B data will record slot_4 and slot_5 as exposed when most users never see them. Report the visible count, or record compact versus expanded mode, using a container-width signal such as useElementSize on the panel root.
Raised by 2 of 6 reviewers (claude-opus-5-5-xhigh adversarial, claude-opus-5-5-xhigh edge-case).
There was a problem hiding this comment.
Fixed in 235cde9. A click now reports how many suggestions were visible at that moment, read from the rendered buttons, so a compact panel reports 3 rather than 5. A unit case covers it and fails without the change.
On exposure: the rendered event carries only the assignment, not slot ids, and both arms hide slots 4 and 5 the same way, so the A/B comparison stays balanced.
| await page.mouse.move(handleCenterX - (width - 420), handleBox.y + 20) | ||
| await page.mouse.up() | ||
| await expect(dock).toHaveCSS('width', `${width}px`) | ||
| await agentPanel.resizeTo(width) |
There was a problem hiding this comment.
🟢 Low — resizeTo(width) sizes and asserts #agent-panel-root, not the outer docked-agent-panel, so this test now measures something different. The shell padding and borders make the dock about 18px wider than before (roughly 498px and 658px instead of 480px and 640px), but the test title still says "panel width". The PM-672 regression no longer exercises the original 480px dock, which is about 462px of content, close to the 460px breakpoint. Either keep asserting the dock width here, or rename and adjust the widths so the test still covers the user-visible case.
Raised by 2 of 6 reviewers (claude-opus-5-5-xhigh adversarial, claude-opus-5-5-xhigh edge-case).
There was a problem hiding this comment.
Fixed in 235cde9. resizeTo now takes the element to size. The X-01 cases pass dockedPanel, so they assert the 480px and 640px dock widths they were written against. The PM-1373 case still measures the panel root, which is what the container query resolves against.
dante01yoon
left a comment
There was a problem hiding this comment.
I went looking for an off-by-one at the breakpoint and didn't find one — the implementation is correct, and the thing I'd have flagged on a Tailwind v3 repo doesn't apply here.
The worry was that @min-[460px] and @max-[460px] both match at exactly 460px, which would hide suggestions 4 and 5 at the width your test asserts them visible. On Tailwind v3 that's real: @max-[460px] compiles to (max-width: 460px), inclusive. This repo is on v4.3.2 (tailwindcss@4.3.2 via the catalog), where the max-* variants use strict range syntax — (width < 460px) against (width >= 460px). No overlap, and your 459/460 pair lands exactly on the two sides of the boundary. Worth knowing that this assertion is load-bearing on the v4 semantics.
Also checked: cn was already imported (EmptyState.vue:5), so the template change is sound, and includeHidden: true really is necessary the way your comment says — the compact state keeps all five in the DOM, so a visible-only locator couldn't distinguish hidden from absent.
Your regression proof (revert EmptyState.vue → fails at suggestions.nth(4) → toBeHidden) is the right evidence for a CSS-only change, and it's the part I'd have asked for if it weren't there.
Two small notes inline. Neither is blocking.
Why Comment rather than Approve — and what would flip it:
Your description says remote checks were still running and you weren't claiming them. They've since resolved, and the state is worth a look: 154 pass, 6 "fail", but all six are cancelled, not failed. gh pr checks renders cancelled as fail, which is misleading here:
CI: Tests E2E created 01:45:51Z cancelled 01:46:09Z
CI: Playwright Videos created 01:45:51Z cancelled 01:45:55Z
Both died within seconds of starting, which reads like concurrency-group cancellation rather than anything in this branch. So there's no failure to chase — but the cancelled set is every Playwright job, including playwright-video-new-tests, the one that would run the case this PR exists to add. The head hasn't moved since (f63a35cc), so as it stands the only evidence the new test passes is your local run.
For a CSS-breakpoint change whose entire deliverable is a browser assertion, I'd rather not sign off on local evidence alone. Re-run those two workflows and if the new case comes back green I'll approve on the spot — I don't expect anything else to come out of it.
| }) => { | ||
| await agentPanel.open() | ||
| const suggestions = agentPanel.suggestedPrompts | ||
| await expect(suggestions).toHaveCount(5) |
There was a problem hiding this comment.
nit-pick (non-blocking): This hardcodes five while the rest of the PR is carefully written to survive the prompt set varying.
Your description makes the point well — the locator resolves by position rather than text precisely because agent.suggestedPrompts became { cloud, local, treatment.cloud }, and the hidden-overflow rule is index-based so it holds for every set. This line is the one place that still assumes a specific set size. On current main both suggestedPrompts.cloud and suggestedPrompts.local are length 5, so it's accurate today; it just couples a breakpoint test to the experiment's set sizes.
It fails loudly rather than silently if a set changes, so this is purely about not having to touch this file when someone adds a sixth starter prompt. toHaveCount(await suggestions.count()) would be circular — but asserting toBeGreaterThan(COMPACT_PROMPT_COUNT) and then checking nth(0..2) visible plus nth(3) hidden would test the same property without pinning the total.
There was a problem hiding this comment.
Done in 235cde9. The case no longer pins five. It reads the rendered count, requires more than the compact count, and checks each suggestion past the third is hidden at 459px and visible at 460px.
| handleY | ||
| ) | ||
| await this.page.mouse.up() | ||
| await expect(this.root).toHaveCSS('width', `${width}px`) |
There was a problem hiding this comment.
nit-pick (non-blocking): Worth calling out in the PR description that this quietly changes what the two existing X-01 cases assert.
They previously measured the dock — expect(dock).toHaveCSS('width', ...) on [data-testid="docked-agent-panel"] — and after the refactor they measure this.root (#agent-panel-root). Your docstring explains why the new test needs the root (it's what the container queries resolve against), and that reasoning is right. But the X-01 cases inherited the switch as a side effect of reusing resizeTo, so they're now pinning a different element than they were written against.
They pass locally per your run, so the two widths evidently agree today. The risk is narrow: if padding or a border is ever introduced between dock and root, the X-01 assertions will start failing for a reason that has nothing to do with what those tests are about, and the trail back to this refactor is easy to lose.
Dropping the measured start width in favour of the hardcoded 420 is a clear improvement, by the way — that's the part of the old inline code most likely to have rotted.
There was a problem hiding this comment.
Addressed in 235cde9 rather than only documented. The X-01 cases measure the dock again via resizeTo(width, agentPanel.dockedPanel), so a future dock/root padding change can't fail them for an unrelated reason.
|
@dante01yoon the Playwright run you asked for is green at Two changes since your review:
One unrelated cloud shard failed once in |
dante01yoon
left a comment
There was a problem hiding this comment.
Re-reviewed at 5e260d95 (was f63a35cc).
Both of my earlier notes are fixed: the PM-1373 case no longer pins five, and the X-01 cases measure the dock again through resizeTo(width, agentPanel.dockedPanel). CI is green at this head (110 pass, 22 skipped), and that was the only thing keeping my last review at Comment.
One new blocking issue came in with 235cde9c7f. prompt_count now carries a different value than the telemetry contract documents, and nothing tells the experiment's readers about it. Details inline. There are also two small notes on the test changes.
| prompt, | ||
| index, | ||
| prompts.value.length, | ||
| visiblePromptCount(), |
There was a problem hiding this comment.
issue: This changes the meaning of an existing analytics field, and the field's own contract now says the opposite.
promptCount becomes prompt_count on app:agent_starter_prompt_clicked (useComposer.ts:75). Its docblock at src/platform/telemetry/types.ts:904 says:
Size of the rendered set, so a set that grew or shrank is visible too.
After this change the same field reports 3 for a click in a narrow panel and 5 for a click in a wide one, with the same prompt set. That is the signal the field exists to give for a set change. It now also moves with panel width, and the starter-prompt set experiment (#20272) reads it by assignment. A dashboard that checks prompt_count to confirm which set rendered will see a mix of 3s and 5s in every arm and can't tell why.
Reporting what the user could see is a reasonable thing to want, but it is a separate fact from the set's size. Either of these works for me:
- Keep
prompt_countasprompts.value.lengthand add a separate field (for examplevisible_prompt_count) for the width-dependent count. - Or, if the analytics owner agrees the field should mean "visible count", update the
types.tsdocblock to say so in this PR and confirm with whoever owns the feat(agent): run starter prompt set experiment #20272 readout, so the change shows up in their data with a known start date.
I'd lean toward the first. It keeps an existing field from changing meaning partway through an experiment.
There was a problem hiding this comment.
Fixed in 30fee969e1, taking your first option.
prompt_count is back to prompts.value.length, so it means "set size" again and the #20272 readout sees a stable 5 per arm. The width-dependent fact moved to a new visible_prompt_count, which reports 3 in a compact panel and the set size elsewhere. The types.ts docblock for prompt_count is unchanged because it is accurate again; the new field carries its own docblock saying why the two are separate, so the next reader does not re-derive this.
The field is required rather than optional on AgentStarterPromptClickedMetadata. That cost six fixture updates, which is the point — no emitter can add a click event and silently omit it.
Coverage, all failing without the change:
EmptyState.test.ts— hides the overflow as the breakpoint does, then assertspromptCount: 5andvisiblePromptCount: 3on the same click. Reverting to the blocked behaviour fails it; unwiring the visible count fails it too.starterPrompts.test.ts— the exact-shape case now pinsvisiblePromptCount, plus a case that the two counts stay independent.AgentPanelRoot.test.ts— the full click→telemetry path asserts both fields on the emitted event.
No analytics owner sign-off is needed for this route, since no existing field changes meaning.
| await expect(agentPanel.suggestedPromptVisibility()).resolves.toEqual([ | ||
| true, | ||
| true, | ||
| true, |
There was a problem hiding this comment.
nit-pick (non-blocking): Two small things in the aggregated assertion.
- The three literal
trues repeatcompactCount. If the compact limit changes, line 12 and this array have to change together....Array.from({ length: compactCount }, () => true)keeps them tied. expect(promise).resolves.toEqual(...)takes one snapshot and doesn't retry, unlike thetoBeVisible/toBeHiddencalls it replaced. I think it's deterministic today, becauseresizeTowaits for the computed width andgetComputedStyleforces the container-query recalc. Still,await expect.poll(() => agentPanel.suggestedPromptVisibility()).toEqual(...)gets back the web-first retry for free if a transition is ever added to the hide.
There was a problem hiding this comment.
Both done in 35f9d67253.
- The literal
trues are derived fromcompactCountnow, so the compact limit lives in one place. - Switched to
await expect.poll(() => agentPanel.suggestedPromptVisibility()).toEqual(...)for both phases, which restores the web-first retry thetoBeVisible/toBeHiddencalls had. You were right that it is deterministic today; the retry is there so adding a transition to the hide cannot turn this flaky later.expect.pollis already the idiom elsewhere inbrowser_tests/fixtures/.
The spec file still passes 6/6 through the harness at the new head, and the PM-1373 case still fails on the parent component.
| global: { plugins: [i18n] } | ||
| }) | ||
| const buttons = screen.getAllByRole('button') | ||
| // jsdom does not evaluate container queries; hide the overflow as the |
There was a problem hiding this comment.
nit-pick (non-blocking): The unit tests run under happy-dom (vite.config.mts:848), not jsdom, so this comment names the wrong environment.
There was a problem hiding this comment.
Fixed in 30fee969e1 — the comment says happy-dom now. You are right: vite.config.mts:843 sets environment: 'happy-dom', and the point of the comment is that the environment does not evaluate container queries, so naming the wrong one undercut it.
Show only the first three cold-start suggestions below the 460px panel container breakpoint, and keep all five in the DOM so the wider layout is unchanged. Uses the complementary @max-[460px] query so no width falls through both arms. Linear: PM-1373, PM-1370
Adds a cloud browser case that drives the panel to 459px and 460px and asserts which cold-start suggestions render, which is the durable pin for PM-1373. Centralizes the resize-handle drag on AgentPanel so X-01 stops duplicating it, and measures the panel root rather than the outer dock because the root is what the container queries resolve against. The suggestion locator resolves by position with includeHidden, so it distinguishes hidden from absent and does not break when the starter-prompt set changes with distribution or experiment variant.
A narrow panel hides suggestions 4 and 5 with a container query, but the click attribution still reported the full set size. Report the number of suggestions that were visible when the user clicked. Restore the X-01 cases to measuring the outer dock, which they were written against, by letting resizeTo take the element to size. The PM-1373 case now checks every suggestion past the compact count instead of pinning five. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…_count `prompt_count` is documented as the size of the rendered set, and the starter-prompt set experiment reads it that way. Reporting the compact panel's visible count through that same field made a narrow panel indistinguishable from a shrunken set in every arm. The visible count is a separate fact, so it travels as a separate field. `prompt_count` goes back to `prompts.value.length`; `visible_prompt_count` carries what the viewer could actually see, and `types.ts` documents the distinction so the next reader does not have to re-derive it. The field is required rather than optional, so no emitter can silently drop it.
Two review nits on the aggregated assertion. The three literal `true`s repeated `compactCount`, so changing the compact limit meant changing two places; they are derived from it now. And `expect(promise).resolves` takes a single snapshot with no web-first retry, unlike the `toBeVisible`/`toBeHidden` calls it replaced — `expect.poll` restores that, so a transition added to the hide cannot make this flaky.
5e260d9 to
35f9d67
Compare
|
Rebased and the blocking finding is fixed. @dante01yoon — I took your first option on One conflict in the rebase, in No approval was spent by the force-push — this PR has never had one, and there was no earlier changes-requested to re-expose. Local validation at the new head: 4,444 unit tests pass across |
dante01yoon
left a comment
There was a problem hiding this comment.
Re-reviewed at 35f9d672532a93d678d25424cb1b0d4ab0aaec36. The blocking analytics issue is fixed: prompt_count remains the rendered set size and visible_prompt_count carries the separate width-dependent value. Both earlier test notes are addressed.
Own verification in an isolated checkout with its own frozen install: 820 tests across the nine changed unit suites, app/browser typechecks, Cloud production build, changed-file type-aware oxlint and formatting passed. ESLint passed its applicable files; its configuration ignores the unit test files.
Against that production build with a real local ComfyUI bootstrap and controlled Cloud/auth/Agent fixtures: all six existing UX scenarios and two additional control/treatment browser probes passed. The probes verify three accessible suggestions at 459px, five at 460px, hidden suggestions excluded from keyboard focus, and actual click telemetry retaining set size 5 while visible count changes 3 → 5 → 3. This is scoped UI/telemetry verification, not a live Cloud account or model execution claim.
Independent frozen disposable-copy controls: conflating set size with visible count failed the narrow-panel unit (1 fail/47 pass); disconnecting visible-count wiring failed the same unit (1 fail/47 pass); restoring head passed 48/48. Removing the CSS hide rule and rebuilding failed the PR's PM-1373 browser test because suggestions four/five remained visible at 459px; the unmodified production build passed that test again.
All five current required checks pass, and their three workflow runs report this exact head. No remaining blocking findings in this diff. No server-state derivation, API response shape, graph/extension callback, or website build change is introduced here.
|
Summary
The other compact-empty-state acceptance item, removing the Comfy logo, is covered by frontend PR 18052, merged 2026-09-25. This PR intentionally avoids duplicating that change.
Rebased onto current
main(second time)5e260d95wasCONFLICTING/DIRTYagain. Rebased ontod965cd4dbd65b24546c518a0319f0f1edb3d4ada; new head35f9d672532a93d678d25424cb1b0d4ab0aaec36. The branch's earliermainmerge commit is gone — the four commits are linear on top ofmainnow.One conflict, in
browser_tests/fixtures/components/AgentPanel.ts.mainadded a privatetoasts: ToastHelperto the property list where this branch addedpublic readonly suggestedPrompts. Both belong, so both are kept, in the list's alphabetical order. Nothing was preferred mechanically and nothing upstream was reverted:main's attachment drop/hand-over helpers, the credit-transition notice locator, the tooltip settle inselectWorkflow,rejectEveryTurn, and the canvas settle in node-selection mode are all intact.EmptyState.vueauto-merged against an upstream class removal on a line this branch does not touch.The net change against
mainis the same four files, +123/−14 before the review fixes below.No approval was spent. This PR has never had one — latest non-dismissed review per reviewer is
dante01yoonCHANGES_REQUESTED,DrJKLCOMMENTED,christian-byrneCOMMENTED.dante01yoonhas no earlier changes-requested to resurrect, so the push could not re-expose a retired verdict.Review follow-up at
35f9d672532a93d678d25424cb1b0d4ab0aaec36The blocking finding was real and is fixed by taking the reviewer's first option.
prompt_countmeant "size of the rendered set" and the starter-prompt set experiment reads it that way; reporting the compact panel's visible count through that same field would have made a narrow panel indistinguishable from a shrunken set in every arm.prompt_countisprompts.value.lengthagain, so its existing docblock is accurate.visible_prompt_countonapp:agent_starter_prompt_clicked: 3 in a compact panel, the set size elsewhere. It carries its own docblock stating why the two are separate.compactCountinstead of repeating three literals, and both breakpoint phases useexpect.pollso they keep a web-first retry. TheEmptyState.test.tscomment names happy-dom, which is the environmentvite.config.mts:843actually configures.Testing
At head
35f9d672532a93d678d25424cb1b0d4ab0aaec36, locally:pnpm vitest run src/platform/telemetry src/workbench/extensions/agent— 209 files, 4444 passed, 1 expected fail.pnpm typecheckandpnpm typecheck:browser— both clean.oxfmt --checkand type-awareoxlintclean on the changed files; full pre-commit hooks ran on both commits, and the pre-pushfallow auditreports no issues in the 15 changed files.@cloudspec file through the browser harness — 6 passed (16.6s), against a local ComfyUI e2e backend withDISTRIBUTION=cloudon the dev server.Regression proofs, all three red without the change:
EmptyState.vuetomainfails the PM-1373 case at the 459px phase — suggestions 4 and 5 resolve visible where the assertion wants them hidden. Restoring it turns the file green.prompt_countcarrying the visible count) fails the newEmptyStatecase onpromptCount: 5.visiblePromptCount()fails the same case onvisiblePromptCount: 3.No existing assertion was weakened to go green.
Remote checks at this head are running; results are not claimed here until they land.
Earlier rebase and review history (head
f63a35cc/235cde9c)Rebased onto current
mainThis branch was
CONFLICTINGagainstmainfor nineteen days and has been rebased onto1697904c9dccce16f50a55d2dfd828bd3aa4e506. New head:f63a35ccbf2f03c8f00665b7290fe4f955c46ed0. The ten review-iteration commits are collapsed into two; the net change is the same three files.Three things moved under the branch and the resolution adapts to them rather than reverting them:
EmptyState.vuegained the starter-prompt set experiment (#20272) and distribution-specific prompt sets (#19283). The click handler is nowonPromptClick(prompt, index), so the class edit is reapplied around it. The hidden-overflow rule is index-based, so it holds for the control set, the local set, and the treatment set alike.agent.suggestedPromptsin the locale files is no longer a flat array — it is{ cloud, local, treatment.cloud }. The page object therefore resolves suggestion buttons by position inside[data-testid="suggested-prompts"], not by prompt text, which would otherwise have broken on whichever set rendered.includeHidden: trueis load-bearing: the compact state keeps every suggestion in the DOM, so a visible-only locator cannot tell "hidden" from "never rendered".main'sX-01 / PM-672already took theagentPanel.open()change but kept the inline resize-handle drag. That duplication is removed here as originally reviewed.Testing
At head
f63a35ccbf2f03c8f00665b7290fe4f955c46ed0, locally:pnpm vitest run src/workbench/extensions/agent/components/agent/EmptyState.test.ts— 24 passedpnpm typecheckandpnpm typecheck:browser— both cleanoxfmt --check,stylelint, and type-awareoxlinton the three changed files — clean. ESLint reports the twobrowser_tests/files as outside its config, which is the repo's configuration, not a change here. Full pre-commit hooks (oxfmt, stylelint, type-aware oxlint, ESLint,typecheck:app) ran on both commits; the transcripts are the recordedlint-stagedoutput.@cloudspec file through the browser harness — 6 passed (21.6s), against a local ComfyUI e2e backend withDISTRIBUTION=cloudon the dev server. That is the file's six cases, i.e. the new breakpoint case plus the two X-01 widths and the three neighbor-layout cases that exercise the refactoredresizeTo.EmptyState.vueand rerunning the new case fails atsuggestions.nth(4)→toBeHidden, resolved to a visible<button>at 459px. Restoring the fix turns it green. The pin fails on the parent and passes on the fix.Remote checks at this head are running; results are not claimed here until they land.
Review follow-up at
235cde9c7ffc790c99ceaeefd30209bd12b13d5aresizeTotakes the element to size; the PM-1373 case still measures the panel root.Locally at this head:
EmptyState.test.ts25 passed;pnpm typecheckandtypecheck:browserclean; the whole@cloudspec file 6 passed. Reverting onlyEmptyState.vuestill fails the new case atnth(3)andnth(4)→toBeHidden.Note for the reviewer: backport labels were retargeted
This PR carried
needs-backport+cloud/1.54. The 1.54 trains are retired, andpr-backport.yamlcreates a carrier per train label on merge — so merging as labelled would have opened acloud/1.54carrier.cloud/1.54was removed and replaced withcloud/1.55+cloud/1.56.Correction, since this paragraph had gone stale:
cloud/1.55was then removed again on 2026-10-11 by the repository owner, so the PR currently carriesneeds-backport+cloud/1.56only. That removal is left alone deliberately — it reads as a release-line decision, not a mechanical slip. If 1.55 should also receive this change, re-add the label before merging. If no release line should, dropneeds-backport.Linear: PM-1370 — compact Agent empty state, PM-1373 — reduce narrow-panel suggestions