-
Notifications
You must be signed in to change notification settings - Fork 584
Expand file tree
/
Copy pathtrend-report.html
More file actions
1019 lines (994 loc) · 29.3 KB
/
Copy pathtrend-report.html
File metadata and controls
1019 lines (994 loc) · 29.3 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>AIDLC Rules Trend Report</title>
<style>:root {
/* AWS Cloudscape-aligned palette */
--aws-squid-ink: #000716;
--aws-orange: #ec7211;
--aws-blue-600: #0972d3;
/* Status colors */
--green-bg: #f2fcf3; --green-text: #037f0c; --green-border: #29ad32;
--yellow-bg: #fff8e1; --yellow-text: #8d6605; --yellow-border: #d4a017;
--red-bg: #fff3f0; --red-text: #d91515; --red-border: #eb5f5f;
--blue-bg: #f0f6ff; --blue-text: #0972d3;
/* Neutral grays */
--gray-50: #fafafa; --gray-100: #f2f3f3; --gray-200: #e9ebed;
--gray-300: #d1d5db; --gray-500: #5f6b7a; --gray-700: #414d5c;
--gray-900: #000716;
--radius: 8px;
}
* { box-sizing: border-box; }
body {
font-family: 'Amazon Ember', -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, 'Helvetica Neue', sans-serif;
max-width: 1200px; margin: 0 auto; padding: 24px;
color: var(--gray-900); background: #fff; line-height: 1.6;
}
h1 { font-size: 28px; margin: 0 0 4px 0; }
h2 {
font-size: 20px; margin: 40px 0 12px 0; padding-bottom: 8px;
border-bottom: 2px solid var(--gray-200); color: var(--gray-900);
}
h3 { font-size: 16px; margin: 24px 0 8px 0; color: var(--gray-700); }
/* Hero header */
.hero {
margin-bottom: 32px; padding: 20px 24px;
background: var(--aws-squid-ink); color: #fff; border-radius: var(--radius);
}
.hero h1 { font-size: 28px; color: #fff; }
.hero .meta { color: #a8b4c4; font-size: 14px; margin-top: 4px; }
/* Navigation */
.nav {
display: flex; flex-wrap: wrap; gap: 6px; margin-bottom: 28px;
padding: 12px 16px; background: var(--aws-squid-ink); border-radius: var(--radius);
border: none;
}
.nav a {
font-size: 13px; color: #d5dbdb; text-decoration: none;
padding: 4px 10px; border-radius: 4px; transition: background 0.15s;
}
.nav a:hover { background: rgba(255, 255, 255, 0.12); color: #fff; }
/* Summary cards */
.cards { display: grid; grid-template-columns: repeat(auto-fit, minmax(170px, 1fr)); gap: 12px; margin: 16px 0 24px 0; }
.card {
padding: 14px 16px; border-radius: var(--radius);
border: 1px solid var(--gray-200); background: #fff;
}
.card .label { font-size: 12px; color: var(--gray-500); text-transform: uppercase; letter-spacing: 0.5px; }
.card .value { font-size: 24px; font-weight: 700; margin: 4px 0 2px 0; }
.card .detail { font-size: 12px; color: var(--gray-500); }
.card.good { border-left: 4px solid var(--green-border); }
.card.warn { border-left: 4px solid var(--yellow-border); }
.card.bad { border-left: 4px solid #d91515; }
/* Tables */
table {
border-collapse: collapse; width: 100%; margin: 12px 0 24px 0;
font-size: 14px; border-radius: var(--radius); overflow: hidden;
border: 1px solid var(--gray-200);
}
th {
background: var(--gray-100); font-weight: 600; text-align: left;
padding: 10px 14px; border-bottom: 2px solid var(--gray-200);
font-size: 13px; color: var(--gray-700); text-transform: uppercase;
letter-spacing: 0.3px;
}
td { padding: 9px 14px; border-bottom: 1px solid var(--gray-100); }
tr:hover td { background: var(--gray-50); }
td:first-child { font-weight: 500; }
/* Score cells */
.s-green { background: var(--green-bg); color: var(--green-text); font-weight: 600; }
.s-yellow { background: var(--yellow-bg); color: var(--yellow-text); font-weight: 500; }
.s-red { background: var(--red-bg); color: var(--red-text); font-weight: 600; }
.d-pos { color: var(--green-text); font-weight: 500; }
.d-neg { color: var(--red-text); font-weight: 500; }
.na { color: var(--gray-500); font-style: italic; }
/* Badges */
.badge {
display: inline-block; padding: 2px 8px; border-radius: 10px;
font-size: 12px; font-weight: 600; letter-spacing: 0.3px;
}
.badge-pass { background: var(--green-bg); color: var(--green-text); }
.badge-fail { background: var(--red-bg); color: var(--red-text); }
.badge-warn { background: var(--yellow-bg); color: var(--yellow-text); }
/* Mini bar chart (CSS only) */
.bar-cell { position: relative; }
.bar {
display: inline-block; height: 16px; border-radius: 2px;
background: linear-gradient(90deg, #ec7211, #ff9900);
vertical-align: middle; margin-right: 6px;
min-width: 2px;
}
/* Two-column split layout */
.split {
display: grid; grid-template-columns: 1fr 2fr;
gap: 24px; align-items: start; margin: 4px 0 24px 0;
}
.split-desc {
font-size: 14px; color: var(--gray-500); line-height: 1.7;
padding-top: 4px;
}
.split-desc p { margin: 0 0 8px 0; }
.split table { margin-top: 0; }
/* Blockquote callouts */
.callout {
padding: 12px 16px; margin: 12px 0;
border-left: 4px solid var(--yellow-border); background: var(--yellow-bg);
border-radius: 0 var(--radius) var(--radius) 0; font-size: 14px;
}
.callout.info { border-left-color: var(--aws-blue-600); background: var(--blue-bg); }
/* Section description */
.section-desc { color: var(--gray-500); font-size: 14px; margin: 0 0 12px 0; }
/* Responsive */
@media (max-width: 768px) {
body { padding: 12px; }
.cards { grid-template-columns: repeat(2, 1fr); }
.split { grid-template-columns: 1fr; }
table { font-size: 13px; }
th, td { padding: 6px 8px; }
}
</style>
</head>
<body>
<div class="hero">
<h1>AIDLC Rules Trend Report</h1>
<div class="meta">6 releases (v0.1.0 through v0.1.5) · awslabs/aidlc-workflows · 2026-03-20T16:04:28.635438+00:00</div>
</div>
<nav class="nav"><a href="#a-executive-summary">A. Summary</a> <a href="#b-functional-correctness">B. Correctness</a> <a href="#c-qualitative-evaluation">C. Qualitative</a> <a href="#d-efficiency-cost-metrics">D. Efficiency</a> <a href="#e-code-quality">E. Quality</a> <a href="#f-stability-reliability">F. Stability</a> <a href="#g-version-over-version-deltas">G. Deltas</a> <a href="#h-pre-release-data-points">H. Pre-Release</a></nav>
<h2 id="a-executive-summary">A. Executive Summary</h2>
<div class="cards">
<div class="card good"><div class="label">Qualitative Score</div><div class="value">0.898</div><div class="detail">Golden: 0.854</div></div>
<div class="card good"><div class="label">Contract Tests</div><div class="value">88/88</div><div class="detail">100.0% pass rate</div></div>
<div class="card good"><div class="label">Unit Tests</div><div class="value">100.0%</div><div class="detail">175/175 passed</div></div>
<div class="card good"><div class="label">Lint Findings</div><div class="value">0</div><div class="detail">Golden: 0</div></div>
<div class="card good"><div class="label">Execution Time</div><div class="value">17.9m</div><div class="detail">Golden: 23.8m</div></div>
<div class="card good"><div class="label">Total Tokens</div><div class="value">13.66M</div><div class="detail">Golden: 18.39M</div></div>
</div>
<p class="section-desc">High-level snapshot comparing the latest release against the golden baseline (the reference evaluation used as the quality target).</p>
<table>
<thead>
<tr>
<th>Metric</th>
<th>What it measures</th>
</tr>
</thead>
<tbody>
<tr><td><strong>Unit test pass rate</strong></td><td>Percentage of generated unit tests that pass. Higher means more reliable code generation.</td></tr>
<tr><td><strong>Contract tests</strong></td><td>API compliance checks against the OpenAPI spec (passed/total). 88/88 = full compliance.</td></tr>
<tr><td><strong>Lint findings</strong></td><td>Static analysis warnings in generated code. Lower is better — 0 means clean code.</td></tr>
<tr><td><strong>Qualitative score</strong></td><td>AI-graded documentation quality on a 0–1 scale (higher is better).</td></tr>
<tr><td><strong>Execution time</strong></td><td>Wall-clock time for the full evaluation run. Lower means faster generation.</td></tr>
<tr><td><strong>Total tokens</strong></td><td>Total LLM tokens consumed (input + output). Lower means more cost-efficient.</td></tr>
</tbody>
</table>
<table>
<thead>
<tr>
<th>Metric</th>
<th>Golden</th>
<th>Latest (v0.1.5)</th>
<th>vs Golden</th>
</tr>
</thead>
<tbody>
<tr>
<td>Unit test pass rate</td>
<td>100.0% (180/180)</td>
<td>100.0% (175/175)</td>
<td>=</td>
</tr>
<tr>
<td>Contract tests</td>
<td>88/88</td>
<td>88/88</td>
<td>=</td>
</tr>
<tr>
<td>Lint findings</td>
<td>0</td>
<td>0</td>
<td>=</td>
</tr>
<tr>
<td>Qualitative score</td>
<td>0.854</td>
<td>0.898</td>
<td class="d-pos">+0.044</td>
</tr>
<tr>
<td>Execution time</td>
<td>23.8m</td>
<td>17.9m</td>
<td class="d-pos">-5.9m</td>
</tr>
<tr>
<td>Total tokens</td>
<td>18.39M</td>
<td>13.66M</td>
<td class="d-pos">-4.74M</td>
</tr>
</tbody>
</table>
<h2 id="b-functional-correctness">B. Functional Correctness</h2>
<p class="section-desc">Measures whether the code generated by each rules version actually works correctly. This is the most fundamental quality gate — code that doesn’t pass its own tests is broken.</p>
<h3>B.1 Unit Tests</h3>
<div class="split">
<div class="split-desc">
<p>Unit tests validate individual functions and components in isolation. The AIDLC rules instruct the AI to generate both source code and test suites.</p>
<p><strong>Pass/Total</strong> = tests that passed out of total generated. <strong>Rate</strong> = pass percentage (100% = all tests passing). <strong>Failures</strong> = tests that ran but produced wrong results.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th>Pass/Total</th>
<th>Rate</th>
<th>Failures</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td>250/250</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.1</td>
<td>194/194</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.2</td>
<td>180/180</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.3</td>
<td>126/126</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.4</td>
<td>156/156</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.5</td>
<td>175/175</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3>B.2 Contract Tests (API Compliance)</h3>
<div class="split">
<div class="split-desc">
<p>Contract tests verify that the generated API implementation matches its OpenAPI specification. Each test sends a request to an endpoint and checks that the HTTP status code and response shape match the spec.</p>
<p>88 endpoints are tested per version. <strong>Pass/Total</strong> = endpoints that returned the expected status code. <strong>Rate</strong> = pass percentage (100% = full spec compliance).</p>
<p><strong>Failures</strong> lists the specific endpoints that deviated from the spec.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th>Pass/Total</th>
<th>Rate</th>
<th>Failures</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td>88/88</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.1</td>
<td>88/88</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.2</td>
<td>88/88</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.3</td>
<td>85/88</td>
<td class="s-green">96.6%</td>
<td class="d-neg">3</td>
</tr>
<tr>
<td>v0.1.4</td>
<td>88/88</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
<tr>
<td>v0.1.5</td>
<td>88/88</td>
<td class="s-green">100.0%</td>
<td>0</td>
</tr>
</tbody>
</table>
</div>
</div>
<div class="callout"><strong>v0.1.3 failures:</strong><ul>
<li><code>POST /api/v1/arithmetic/add</code> — expected 422, got 200 (add missing field → 422)</li>
<li><code>POST /api/v1/arithmetic/divide</code> — expected 400, got 200 (divide by zero → error)</li>
<li><code>POST /api/v1/arithmetic/modulo</code> — expected 400, got 200 (modulo by zero → error)</li>
</ul></div>
<h2 id="c-qualitative-evaluation">C. Qualitative Evaluation</h2>
<p class="section-desc">Measures the quality of generated documentation by comparing it against human-authored reference documents. An AI evaluator scores each document on completeness, accuracy, and clarity, producing a 0–1 score (1.0 = perfect match to reference quality).</p>
<h3>C.1 Overall Score</h3>
<div class="split">
<div class="split-desc">
<p>The weighted average across all evaluated documents. This is the single best indicator of how well the rules produce documentation.</p>
<p>Scores above 0.90 are considered strong; below 0.70 signals significant gaps.</p>
<p>Golden baseline: <strong>0.854</strong></p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th>Overall</th>
<th>vs Golden</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td class="s-yellow">0.860</td>
<td class="d-pos">+0.006</td>
</tr>
<tr>
<td>v0.1.1</td>
<td class="s-yellow">0.888</td>
<td class="d-pos">+0.033</td>
</tr>
<tr>
<td>v0.1.2</td>
<td class="s-yellow">0.893</td>
<td class="d-pos">+0.038</td>
</tr>
<tr>
<td>v0.1.3</td>
<td class="s-yellow">0.866</td>
<td class="d-pos">+0.012</td>
</tr>
<tr>
<td>v0.1.4</td>
<td class="s-yellow">0.891</td>
<td class="d-pos">+0.037</td>
</tr>
<tr>
<td>v0.1.5</td>
<td class="s-yellow">0.898</td>
<td class="d-pos">+0.044</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3>C.2 Phase Breakdown</h3>
<div class="split">
<div class="split-desc">
<p>Documents are grouped by SDLC phase. <strong>Inception</strong> covers early-stage design artifacts (requirements, architecture plans, component designs) — these are generated first and set the foundation.</p>
<p><strong>Construction</strong> covers build-time artifacts (build instructions, test instructions, build-and-test summaries) — these depend on inception outputs being correct.</p>
<p>A drop in inception quality often cascades into construction.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th>Inception</th>
<th>Construction</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td class="s-yellow">0.880</td>
<td class="s-yellow">0.840</td>
</tr>
<tr>
<td>v0.1.1</td>
<td class="s-yellow">0.894</td>
<td class="s-yellow">0.882</td>
</tr>
<tr>
<td>v0.1.2</td>
<td class="s-green">0.921</td>
<td class="s-yellow">0.864</td>
</tr>
<tr>
<td>v0.1.3</td>
<td class="s-yellow">0.886</td>
<td class="s-yellow">0.846</td>
</tr>
<tr>
<td>v0.1.4</td>
<td class="s-yellow">0.890</td>
<td class="s-yellow">0.892</td>
</tr>
<tr>
<td>v0.1.5</td>
<td class="s-yellow">0.879</td>
<td class="s-green">0.918</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3>C.3 Per-Document Heatmap</h3>
<p class="section-desc">Individual quality scores for each generated document across all versions. This reveals which specific documents are consistently strong, improving, or problematic. Documents scoring below 0.70 (red) are the top candidates for rules improvements.</p>
<table>
<thead>
<tr>
<th>Document</th>
<th>v0.1.0</th>
<th>v0.1.1</th>
<th>v0.1.2</th>
<th>v0.1.3</th>
<th>v0.1.4</th>
<th>v0.1.5</th>
</tr>
</thead>
<tbody>
<tr>
<td><code>application-design-plan.md</code></td>
<td><span class="na">—</span></td>
<td class="s-green">0.96</td>
<td class="s-green">1.00</td>
<td class="s-green">0.96</td>
<td class="s-green">1.00</td>
<td class="s-green">0.95</td>
</tr>
<tr>
<td><code>build-and-test-summary.md</code></td>
<td class="s-green">0.93</td>
<td class="s-green">0.95</td>
<td class="s-green">0.90</td>
<td class="s-green">0.90</td>
<td class="s-green">0.97</td>
<td class="s-green">0.95</td>
</tr>
<tr>
<td><code>build-instructions.md</code></td>
<td class="s-yellow">0.75</td>
<td class="s-yellow">0.75</td>
<td class="s-yellow">0.78</td>
<td class="s-yellow">0.88</td>
<td class="s-yellow">0.77</td>
<td class="s-yellow">0.87</td>
</tr>
<tr>
<td><code>component-dependency.md</code></td>
<td class="s-green">0.97</td>
<td class="s-green">0.95</td>
<td class="s-green">0.96</td>
<td class="s-green">0.96</td>
<td class="s-green">1.00</td>
<td class="s-green">0.95</td>
</tr>
<tr>
<td><code>component-methods.md</code></td>
<td class="s-green">0.93</td>
<td class="s-green">0.90</td>
<td class="s-green">0.96</td>
<td class="s-green">0.98</td>
<td class="s-green">0.93</td>
<td class="s-green">0.96</td>
</tr>
<tr>
<td><code>components.md</code></td>
<td class="s-green">1.00</td>
<td class="s-green">0.98</td>
<td class="s-green">1.00</td>
<td class="s-green">0.97</td>
<td class="s-green">0.98</td>
<td class="s-green">0.98</td>
</tr>
<tr>
<td><code>execution-plan.md</code></td>
<td class="s-green">0.97</td>
<td class="s-green">0.91</td>
<td class="s-green">0.98</td>
<td class="s-green">0.93</td>
<td class="s-green">0.97</td>
<td class="s-green">0.97</td>
</tr>
<tr>
<td><code>integration-test-instructions.md</code></td>
<td class="s-yellow">0.85</td>
<td class="s-yellow">0.87</td>
<td class="s-yellow">0.82</td>
<td class="s-yellow">0.70</td>
<td class="s-yellow">0.88</td>
<td class="s-green">0.91</td>
</tr>
<tr>
<td><code>requirement-verification-questions.md</code></td>
<td class="s-red">0.38</td>
<td class="s-red">0.54</td>
<td class="s-red">0.54</td>
<td class="s-red">0.38</td>
<td class="s-red">0.36</td>
<td class="s-red">0.28</td>
</tr>
<tr>
<td><code>requirements.md</code></td>
<td class="s-green">1.00</td>
<td class="s-green">1.00</td>
<td class="s-green">0.97</td>
<td class="s-green">1.00</td>
<td class="s-green">0.97</td>
<td class="s-green">0.97</td>
</tr>
<tr>
<td><code>sci-calc-code-generation-plan.md</code></td>
<td class="s-green">0.97</td>
<td class="s-green">0.98</td>
<td class="s-green">0.92</td>
<td class="s-green">0.98</td>
<td class="s-green">0.98</td>
<td class="s-green">0.98</td>
</tr>
<tr>
<td><code>services.md</code></td>
<td class="s-green">0.91</td>
<td class="s-green">0.91</td>
<td class="s-green">0.96</td>
<td class="s-green">0.91</td>
<td class="s-green">0.91</td>
<td class="s-green">0.97</td>
</tr>
<tr>
<td><code>unit-test-instructions.md</code></td>
<td class="s-yellow">0.70</td>
<td class="s-yellow">0.86</td>
<td class="s-green">0.90</td>
<td class="s-yellow">0.77</td>
<td class="s-yellow">0.86</td>
<td class="s-yellow">0.88</td>
</tr>
</tbody>
</table>
<p class="section-desc"><span class="badge badge-pass">green ≥ 0.90</span> <span class="badge badge-warn">yellow 0.70–0.89</span> <span class="badge badge-fail">red < 0.70</span></p>
<h3>C.4 Document Coverage</h3>
<p class="section-desc">Tracks whether the generated output includes the same set of documents as the reference. <strong>Unmatched Ref</strong> = reference documents the AI failed to generate (missing output). <strong>Unmatched Candidate</strong> = extra documents the AI generated that don’t exist in the reference (unexpected output). Ideally both columns are 0, meaning the AI produced exactly the expected set of documents.</p>
<table>
<thead>
<tr>
<th>Version</th>
<th>Unmatched Ref</th>
<th>Unmatched Candidate</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td class="d-neg">1</td>
<td class="d-neg">1</td>
</tr>
<tr>
<td>v0.1.1</td>
<td>0</td>
<td class="d-neg">4</td>
</tr>
<tr>
<td>v0.1.2</td>
<td>0</td>
<td class="d-neg">1</td>
</tr>
<tr>
<td>v0.1.3</td>
<td>0</td>
<td class="d-neg">6</td>
</tr>
<tr>
<td>v0.1.4</td>
<td>0</td>
<td class="d-neg">6</td>
</tr>
<tr>
<td>v0.1.5</td>
<td>0</td>
<td>0</td>
</tr>
</tbody>
</table>
<h2 id="d-efficiency-cost-metrics">D. Efficiency & Cost Metrics</h2>
<p class="section-desc">Tracks the computational resources consumed by each evaluation run. These metrics directly affect cost (tokens) and developer wait time (execution time). Lower values are generally better, as long as quality metrics remain stable.</p>
<h3>D.1 Token Consumption</h3>
<div class="split">
<div class="split-desc">
<p>Total LLM tokens consumed during the run, broken down by agent. <strong>Total</strong> = all tokens across all agents (input + output).</p>
<p><strong>Executor</strong> = the agent that generates code and documents. <strong>Simulator</strong> = the agent that simulates user interactions for testing.</p>
<p>Token count is the primary cost driver — each token represents a unit of LLM usage billed by the provider.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th></th>
<th>Total</th>
<th>Executor</th>
<th>Simulator</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td class="bar-cell"><span class="bar" style="width:68%"></span></td>
<td>9.26M</td>
<td>4.65M</td>
<td>119.3K</td>
</tr>
<tr>
<td>v0.1.1</td>
<td class="bar-cell"><span class="bar" style="width:98%"></span></td>
<td>13.34M</td>
<td>6.56M</td>
<td>266.2K</td>
</tr>
<tr>
<td>v0.1.2</td>
<td class="bar-cell"><span class="bar" style="width:61%"></span></td>
<td>8.34M</td>
<td>4.15M</td>
<td>295.5K</td>
</tr>
<tr>
<td>v0.1.3</td>
<td class="bar-cell"><span class="bar" style="width:84%"></span></td>
<td>11.52M</td>
<td>5.72M</td>
<td>222.3K</td>
</tr>
<tr>
<td>v0.1.4</td>
<td class="bar-cell"><span class="bar" style="width:84%"></span></td>
<td>11.52M</td>
<td>5.67M</td>
<td>251.9K</td>
</tr>
<tr>
<td>v0.1.5</td>
<td class="bar-cell"><span class="bar" style="width:100%"></span></td>
<td>13.66M</td>
<td>6.88M</td>
<td>90.2K</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3>D.2 Execution Time</h3>
<div class="split">
<div class="split-desc">
<p>Wall-clock duration of the full evaluation pipeline, broken down by handoff. Each <strong>handoff</strong> (H1, H2, H3) represents a sequential phase.</p>
<p>H1 is typically code generation (the longest phase), H2 is build/test execution, and H3 is result collection and reporting.</p>
<p><strong>Wall Clock</strong> is the total end-to-end time.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th></th>
<th>Wall Clock</th>
<th>Handoff Breakdown</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td class="bar-cell"><span class="bar" style="width:85%"></span></td>
<td>16.0m</td>
<td>H1: 13.8m · H2: 0.9m · H3: 1.3m</td>
</tr>
<tr>
<td>v0.1.1</td>
<td class="bar-cell"><span class="bar" style="width:99%"></span></td>
<td>18.6m</td>
<td>H1: 17.1m · H2: 1.0m · H3: 0.5m</td>
</tr>
<tr>
<td>v0.1.2</td>
<td class="bar-cell"><span class="bar" style="width:82%"></span></td>
<td>15.5m</td>
<td>H1: 11.7m · H2: 1.4m · H3: 2.4m</td>
</tr>
<tr>
<td>v0.1.3</td>
<td class="bar-cell"><span class="bar" style="width:100%"></span></td>
<td>18.8m</td>
<td>H1: 15.8m · H2: 1.3m · H3: 1.7m</td>
</tr>
<tr>
<td>v0.1.4</td>
<td class="bar-cell"><span class="bar" style="width:89%"></span></td>
<td>16.8m</td>
<td>H1: 14.8m · H2: 1.3m · H3: 0.6m</td>
</tr>
<tr>
<td>v0.1.5</td>
<td class="bar-cell"><span class="bar" style="width:95%"></span></td>
<td>17.9m</td>
<td>H1: 15.0m · H2: 0.7m · H3: 2.2m</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3>D.3 Context Window Pressure</h3>
<div class="split">
<div class="split-desc">
<p>Measures how much of the LLM’s context window is being used across API calls. <strong>Max</strong> = the largest single context seen during the run (approaching the model’s limit risks truncation or degraded output).</p>
<p><strong>Avg</strong> = the mean context size across all API calls. <strong>Median</strong> = the midpoint context size (less affected by outliers than avg).</p>
<p>High context pressure can indicate overly verbose prompts or accumulated conversation history.</p>
</div>
<div>
<table>
<thead>
<tr>
<th>Version</th>
<th>Max</th>
<th>Avg</th>
<th>Median</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td>97.4K</td>
<td>44.9K</td>
<td>43.7K</td>
</tr>
<tr>
<td>v0.1.1</td>
<td>138.5K</td>
<td>57.2K</td>
<td>50.4K</td>
</tr>
<tr>
<td>v0.1.2</td>
<td>96.4K</td>
<td>38.8K</td>
<td>26.6K</td>
</tr>
<tr>
<td>v0.1.3</td>
<td>118.6K</td>
<td>49.7K</td>
<td>42.4K</td>
</tr>
<tr>
<td>v0.1.4</td>
<td>109.8K</td>
<td>48.8K</td>
<td>48.5K</td>
</tr>
<tr>
<td>v0.1.5</td>
<td>121.7K</td>
<td>56.6K</td>
<td>55.2K</td>
</tr>
</tbody>
</table>
</div>
</div>
<h2 id="e-code-quality">E. Code Quality</h2>
<p class="section-desc">Static analysis of the generated codebase. These metrics reflect the cleanliness and maintainability of the AI-generated code, independent of whether it passes tests.</p>
<table>
<thead>
<tr>
<th>Metric</th>
<th>What it measures</th>
</tr>
</thead>
<tbody>
<tr><td><strong>Lint Findings</strong></td><td>Warnings from static analysis (style violations, unused variables, etc.). 0 = clean.</td></tr>
<tr><td><strong>Security Findings</strong></td><td>Vulnerabilities detected by security scanners (SQL injection, XSS, etc.). N/A if no scanner was configured.</td></tr>
<tr><td><strong>Source Files</strong></td><td>Number of non-test source files in the generated project.</td></tr>
<tr><td><strong>LOC</strong></td><td>Total lines of code across all source files. Large swings may indicate generated boilerplate or missing modules.</td></tr>
</tbody>
</table>
<table>
<thead>
<tr>
<th>Version</th>
<th>Lint Findings</th>
<th>Security Findings</th>
<th>Source Files</th>
<th>LOC</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>977</td>
<td>398.5K</td>
</tr>
<tr>
<td>v0.1.1</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>977</td>
<td>398.1K</td>
</tr>
<tr>
<td>v0.1.2</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>976</td>
<td>397.4K</td>
</tr>
<tr>
<td>v0.1.3</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>977</td>
<td>397.7K</td>
</tr>
<tr>
<td>v0.1.4</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>976</td>
<td>397.5K</td>
</tr>
<tr>
<td>v0.1.5</td>
<td>0</td>
<td><span class="na">N/A</span></td>
<td>976</td>
<td>397.6K</td>
</tr>
</tbody>
</table>
<h2 id="f-stability-reliability">F. Stability & Reliability</h2>
<p class="section-desc">Tracks whether the evaluation pipeline itself ran smoothly, independent of output quality.</p>
<table>
<thead>
<tr>
<th>Metric</th>
<th>What it measures</th>
</tr>
</thead>
<tbody>
<tr><td><strong>Error Events</strong></td><td>Runtime errors logged during the run (exceptions, timeouts, API failures). 0 = clean run.</td></tr>
<tr><td><strong>Handoffs</strong></td><td>Number of sequential pipeline phases completed. Typically 3 (generate, build/test, report). A different count may indicate an early abort or retry.</td></tr>
<tr><td><strong>Server Startup</strong></td><td>Whether the generated application server started successfully. A failure here means the generated code couldn’t even boot, preventing contract tests from running.</td></tr>
</tbody>
</table>
<table>
<thead>
<tr>
<th>Version</th>
<th>Error Events</th>
<th>Handoffs</th>
<th>Server Startup</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
<tr>
<td>v0.1.1</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
<tr>
<td>v0.1.2</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
<tr>
<td>v0.1.3</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
<tr>
<td>v0.1.4</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
<tr>
<td>v0.1.5</td>
<td>0</td>
<td>3</td>
<td><span class="badge badge-pass">PASS</span></td>
</tr>
</tbody>
</table>
<h2 id="g-version-over-version-deltas">G. Version-over-Version Deltas</h2>
<p class="section-desc">Each row shows the change from one release to the next, making it easy to spot which specific version introduced an improvement or regression. Positive values (+) indicate an increase; negative (−) indicate a decrease. For <strong>Unit Tests</strong> and <strong>Contract</strong>, positive is better (more tests passing). For <strong>Qualitative</strong>, positive is better (higher quality score). For <strong>Tokens</strong> and <strong>Time</strong>, negative is better (more efficient).</p>
<table>
<thead>
<tr>
<th>Transition</th>
<th>Unit Tests</th>
<th>Contract</th>
<th>Qualitative</th>
<th>Tokens</th>
<th>Time</th>
</tr>
</thead>
<tbody>
<tr>
<td>v0.1.0 → v0.1.1</td>
<td class="d-neg">-56</td>
<td>+0</td>
<td class="d-pos">+0.028</td>
<td class="d-neg">+4.08M</td>
<td class="d-neg">+155s</td>
</tr>
<tr>
<td>v0.1.1 → v0.1.2</td>
<td class="d-neg">-14</td>
<td>+0</td>
<td class="d-pos">+0.005</td>
<td class="d-pos">-5.00M</td>
<td class="d-pos">-188s</td>
</tr>
<tr>
<td>v0.1.2 → v0.1.3</td>
<td class="d-neg">-54</td>
<td class="d-neg">-3</td>
<td class="d-neg">-0.026</td>
<td class="d-neg">+3.19M</td>
<td class="d-neg">+200s</td>
</tr>
<tr>
<td>v0.1.3 → v0.1.4</td>
<td class="d-pos">+30</td>
<td class="d-pos">+3</td>
<td class="d-pos">+0.025</td>
<td class="d-pos">-9.2K</td>