Repository navigation
Expand file tree
/
Copy pathStatus.html
More file actions
7719 lines (7719 loc) · 225 KB
/
Copy pathStatus.html
File metadata and controls
7719 lines (7719 loc) · 225 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 xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8" />
<meta name="generator" content="pandoc 3.10" />
<meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes" />
<title>Status</title>
<style>
/* Default styles provided by pandoc.
** See https://pandoc.org/MANUAL.html#variables-for-html for config info.
*/
html {
color: #1a1a1a;
background-color: #fdfdfd;
}
body {
margin: 0 auto;
max-width: 36em;
padding-left: 50px;
padding-right: 50px;
padding-top: 50px;
padding-bottom: 50px;
hyphens: auto;
overflow-wrap: break-word;
text-rendering: optimizeLegibility;
font-kerning: normal;
}
@media (max-width: 600px) {
body {
font-size: 0.9em;
padding: 12px;
}
h1 {
font-size: 1.8em;
}
}
@media print {
html {
background-color: white;
}
body {
background-color: transparent;
color: black;
font-size: 12pt;
}
p, h2, h3 {
orphans: 3;
widows: 3;
}
h2, h3, h4 {
page-break-after: avoid;
}
}
p {
margin: 1em 0;
}
a {
color: #1a1a1a;
}
a:visited {
color: #1a1a1a;
}
img {
max-width: 100%;
}
svg {
height: auto;
max-width: 100%;
}
h1, h2, h3, h4, h5, h6 {
margin-top: 1.4em;
}
h5, h6 {
font-size: 1em;
font-style: italic;
}
h6 {
font-weight: normal;
}
ol, ul {
padding-left: 1.7em;
margin-top: 1em;
}
li > ol, li > ul {
margin-top: 0;
}
blockquote {
margin: 1em 0 1em 1.7em;
padding-left: 1em;
border-left: 2px solid #e6e6e6;
color: #606060;
}
code {
white-space: pre-wrap;
font-family: Menlo, Monaco, Consolas, 'Lucida Console', monospace;
font-size: 85%;
margin: 0;
hyphens: manual;
}
pre {
margin: 1em 0;
overflow: auto;
}
pre code {
padding: 0;
overflow: visible;
overflow-wrap: normal;
}
.sourceCode {
background-color: transparent;
overflow: visible;
}
hr {
border: none;
border-top: 1px solid #1a1a1a;
height: 1px;
margin: 1em 0;
}
table {
margin: 1em 0;
border-collapse: collapse;
width: 100%;
overflow-x: auto;
display: block;
font-variant-numeric: lining-nums tabular-nums;
}
table caption {
margin-bottom: 0.75em;
}
tbody {
margin-top: 0.5em;
border-top: 1px solid #1a1a1a;
border-bottom: 1px solid #1a1a1a;
}
th {
border-top: 1px solid #1a1a1a;
padding: 0.25em 0.5em 0.25em 0.5em;
}
td {
padding: 0.125em 0.5em 0.25em 0.5em;
}
header {
margin-bottom: 4em;
text-align: center;
}
#TOC li {
list-style: none;
}
#TOC ul {
padding-left: 1.3em;
}
#TOC > ul {
padding-left: 0;
}
#TOC a:not(:hover) {
text-decoration: none;
}
span.smallcaps{font-variant: small-caps;}
div.columns{display: flex; gap: 1.5em;}
div.column{flex: auto;}
@media screen {
div.columns{gap: min(4vw, 1.5em);}
div.column{overflow-x: auto;}
}
div.hanging-indent{margin-left: 1.5em; text-indent: -1.5em;}
/* The extra [class] is a hack that increases specificity enough to
override a similar rule in reveal.js */
ul.task-list[class]{list-style: none;}
ul.task-list li input[type="checkbox"] {
font-size: inherit;
width: 0.8em;
margin: 0 0.8em 0.2em -1.6em;
vertical-align: middle;
}
.display.math{display: block; text-align: center; margin: 0.5rem auto;}
</style>
</head>
<body>
<header id="title-block-header">
<h1 class="title">Status</h1>
</header>
<h1 id="helixtranslate--feature-status">HelixTranslate — Feature
Status</h1>
<p><strong>Revision:</strong> 18 <strong>Last modified:</strong>
2026-06-16T20:05:00Z <strong>Authority:</strong> Derived from
<code>docs/features/.feature_inventory_raw.md</code> (Revision 1). Per
§11.4.45 (captured-evidence Status doc), §11.4.44 (revision header),
§11.4.56 (two-audience summary → <code>Status_Summary.md</code>),
§11.4.60 (always-sync), §11.4.6 (no-guessing — every status reflects
what the inventory actually found in source). <strong>Scope:</strong>
Every one of the 417 inventoried features across CLI, API, Web, gRPC,
Library, Submodule and Infra. <code>helix_qa</code> deliberately not
included.</p>
<blockquote>
<p><strong>Anti-bluff note (§11.4.6 / §11.4.2 / §11.4.107):</strong>
<code>Video-confirmation</code> is set to a real file
<strong>only</strong> where a recording exists AND that recording
genuinely demonstrates the feature working (each cited <code>.mp4</code>
was ffprobe-verified for non-zero duration+frames and content-verified
against its claimed behaviour — e.g. real DeepSeek output in the target
language, real gRPC translated text, live WebSocket count change).
<strong>66 feature rows are video-confirmed</strong> (Wave 9 / Wave 8 /
Wave 7 of 2026-06-16 added <strong>12 net-new</strong> (4 Wave 7 + 5
Wave 8 + 3 Wave 9), each a real input → real LLM (default in-process
bridge, NOT <code>-use-verifier</code>) → real output clip,
window-scoped (§11.4.154), <code>helix_translate-</code>-prefixed
(§11.4.155), frame-content-read so source-frame ≠ output-frame per
§11.4.107. <strong>Wave 7 (4):</strong> <strong>EPUB→EPUB round-trip
EN→ES</strong> ("The old man walked along the shore at dawn…" → "El
anciano caminaba por la orilla al amanecer…", output
<code>EPUB document</code> —
<code>helix_translate-epub-to-epub-en-es-wave7-20260616-181058.mp4</code>;
<strong>headline re-confirm</strong> — the existing EPUB→EPUB-conversion
row was already Confirmed, this clip is an additional distinct-scope
confirmation, NOT a new row), <strong>standalone <code>-script</code>
Cyrillic→Latin</strong> (pkg/script, NO LLM: "Добар дан свима…ђ…џ" →
"Dobar dan svima…đ…dž", zero Cyrillic remaining —
<code>helix_translate-script-cyrillic-to-latin-wave7-20260616-181012.mp4</code>;
<strong>headline-only</strong> re-confirm of an already-Confirmed row),
<strong>preparation-translator analysis pass</strong> (genre "science
fiction", chars Sarah Mills/Admiral Chen, untranslatable Captain Sarah
Mills/Admiral Chen/starship Aurora/New Geneva —
<code>helix_translate-preparation-analysis-en-es-wave7-20260616-181123.mp4</code>;
<strong>headline-only</strong> re-confirm — the preparation-translator
row is already Confirmed) and <strong>REST
<code>POST /api/v1/translate/string</code></strong> (local
<code>./build/server</code> :8080, JSON <code>translated_text</code> =
"Las montañas estaban cubiertas de nieve y el río fluía tranquilamente
por el valle.", provider deepseek — a real translation, NOT a session_id
—
<code>helix_translate-rest-translate-string-en-es-wave7-20260616-181155.mp4</code>;
<strong>ROW-FLIPPED</strong> PENDING→Confirmed). <strong>§11.4.138 /
§11.4.146 honest finding:</strong> Wave 7 ALSO recorded the
<strong>markdown-translator EPUB→MD empty-payload data-loss BUG</strong>
(step-3 translate sends empty strings to DeepSeek, the LLM's "you sent
an empty message" boilerplate written AS IF the translation, chapter
text lost) — <strong>found broken (wave7) → fixed at source in commit
4fbe581 (this session's PART-1 Translate-arg data-loss fix) → a
re-record is owed before any confirmation</strong>; it was NOT a
previously-confirmed row so there is NO demotion, and it is NOT counted
as a confirmed net-new. <strong>Wave 8 (5):</strong> <strong>GROQ
provider path EN→ES</strong> (groq/llama-3.3-70b-versatile, real Groq
API: "La vieja farola se erguía sola sobre el acantilado rocoso…" —
<code>helix_translate-groq-en-es-wave8-20260616_182021.mp4</code>;
<strong>ROW-FLIPPED</strong> the <code>Groq client</code> library row
PENDING→Confirmed), and 4 language-pair confirmations through the
deepseek path (<strong>headline-only</strong> — covered by the
already-Confirmed generic multi-language-pair coverage, no fabricated
per-pair rows): <strong>EN→KO Korean/Hangul</strong> ("새벽에 용이 잠든
숨겨진 성을 찾기 위해 용감한 기사가 고대 숲을 가로질렀다." —
<code>helix_translate-en-ko-korean-wave8-20260616_182059.mp4</code>),
<strong>DE→EN reverse</strong> ("The scientist worked all night in his
lab…" —
<code>helix_translate-de-en-reverse-wave8-20260616_182101.mp4</code>),
<strong>FR→ES non-English↔︎non-English</strong> ("El anciano marinero
contaba historias de tormentas y tesoros ocultos…" —
<code>helix_translate-fr-es-pair-wave8-20260616_182111.mp4</code>) and
<strong>EN→SR Latin script</strong> ("U srcu planina živela je mudra
stara žena…", all-Latin, no Cyrillic —
<code>helix_translate-en-sr-latin-wave8-20260616_182113.mp4</code>).
§11.4.6 nit (not a bug): <code>-h</code> provider help omits groq though
groq IS fully wired + works end-to-end. <strong>Wave 9 (3):</strong>
<strong>REST <code>POST /api/v1/translate/batch</code></strong> (live
<code>server-w9</code> :8091 TLS, 3 EN→SR in one request, 0 errors,
provider llm-novita —
<code>helix_translate-rest-batch-version-wave9-20260616_184823.mp4</code>;
<strong>ROW-FLIPPED</strong> PENDING→Confirmed), <strong>REST
<code>GET /api/v1/version</code></strong> (real substantive data:
codebase 2.3.0, git_commit c2aa7c8…, go1.26.2, build_time + components
map, NOT a bare status — same clip; <strong>headline re-confirm</strong>
of the already-Confirmed combined endpoints row) and
<strong>preparation→translate two-stage pipeline</strong> (stage-1 real
DeepSeek analysis: "short story"/"literary fiction", chars Captain
Marlowe + Eli, 3 untranslatable Stormcrag Point/Captain Marlowe/Eli,
1974 tokens; stage-2 output EPUB preserves exactly those names in
Spanish "…El viejo capitán Marlowe había sido farero de Stormcrag Point
durante cuarenta años…" —
<code>helix_translate-prep-translate-two-stage-wave9-20260616_184823.mp4</code>;
<strong>headline-only</strong> — covered by the already-Confirmed
preparation-translator row). <strong>§11.4.6 / §11.4.124 honest
non-confirmable findings (recorded as gaps, NOT confirmed, never
faked):</strong> <code>pkg/api/server.go</code> alt <code>Server</code>
(<code>/api/translate</code>, <code>/api/batch</code>,
<code>/api/languages</code>) is <strong>UNWIRED</strong> — no
<code>cmd</code> entry instantiates <code>api.NewServer</code>, no
runnable binary (did NOT fabricate a runner); and
<code>GET /api/v1/providers</code> returns a <strong>static hardcoded
list</strong> (openai/anthropic/zhipu/deepseek), NOT substantive real
upstream data, so NOT confirmed as substantive per the Wave-9 criterion.
<strong>[Both addressed post-Wave-9, this session:</strong> the
<code>pkg/api/server.go</code> alt <code>Server</code> is now documented
as honest §11.4.124 NEVER-WIRED test-only-scaffolding
<strong>N/A</strong> (see the dedicated section header note below,
kept-not-deleted per operator review); and the
<code>/api/v1/providers</code> static-list was <strong>FIXED in commit
<code>388a2eb</code></strong> — it now serves the real
configured/available provider set (§11.4.115 RED→GREEN + §11.4.135 guard
<code>HTQ-FIX-006</code>), with a live re-record of
<code>/api/v1/providers</code> against the rebuilt nezha image owed
before flipping that aspect to a fresh Confirmed.]** <strong>Running
tally (§11.4.6 honest arithmetic): 54 (prior authoritative total) + 4
(Wave 7) + 5 (Wave 8) + 3 (Wave 9) = 66.</strong> Of the 12 net-new,
<strong>3 are genuine row-flips</strong> (PENDING→Confirmed:
<code>POST /api/v1/translate/string</code>, the <code>Groq client</code>
library row, <code>POST /api/v1/translate/batch</code>); the remaining 9
are headline-counted additional distinct-scope confirmations on rows
that were already Confirmed (EPUB→EPUB round-trip, <code>-script</code>
Cyrillic→Latin, preparation analysis, <code>GET /api/v1/version</code>,
preparation→translate two-stage) or language-pair/provider confirmations
covered by already-Confirmed generic rows (EN→KO, DE→EN, FR→ES,
EN→SR-latin) — counted in the operator-sanctioned running total but
honestly distinguished here so no row is double-flipped. Wave 6 of
2026-06-16 added <strong>5 net-new</strong> via the default in-process
direct DeepSeek CLI path, each ffprobe+frame-verified per §11.4.107
(790×560 yuv420p, advancing frames, English input-frame ≠ translated
output-frame): <strong>HTML→HTML round-trip EN→ES</strong>
(<code>helix_translate-html_html_roundtrip-wave6-20260616-175612.mp4</code>,
real Spanish HTML), <strong>PDF input accepted EN→ES</strong>
(<code>helix_translate-pdf_input-wave6-20260616-175612.mp4</code>, from
a real PDF 1.7), <strong>DOCX input accepted EN→ES</strong>
(<code>helix_translate-docx_input-wave6-20260616-175612.mp4</code>, from
a real Word 2007+ file), <strong>EN→ZH (Chinese)</strong>
(<code>helix_translate-en_zh_translate-wave6-20260616-175612.mp4</code>)
and <strong>EN→RU (Russian)</strong>
(<code>helix_translate-en_ru_translate-wave6-20260616-175612.mp4</code>);
Wave 6 also recorded an honest MINOR finding (HTML/EPUB chapter-title
duplication on output, PENDING_FORENSICS, cosmetic — core translation
correct). Wave 5 of 2026-06-16 added <strong>4 net-new</strong> (the
<code>-multipass</code> row was NOT counted — see §11.4.138 demotion
below): <strong>EN→PT (Portuguese)</strong>
(<code>helix_translate-en-pt-translate-wave5-20260616-173814.mp4</code>),
<strong>EN→JA (Japanese, non-Latin)</strong>
(<code>helix_translate-en-ja-translate-wave5-20260616-173814.mp4</code>),
<strong>REST <code>POST /api/v1/convert/script</code></strong>
(<code>helix_translate-rest-convert-script-wave5-20260616-173814.mp4</code>,
real Cyrillic↔︎Latin JSON both directions) and <strong>REST
<code>POST /api/v1/translate/fb2</code></strong>
(<code>helix_translate-rest-translate-fb2-wave5-20260616-173814.mp4</code>,
FB2→real Serbian EPUB; note: handler hardcodes ru→sr — tracked fix).
Wave 4 of 2026-06-16 added <strong>3 net-new</strong>
multi-language-pair translations from the same EN source proving genuine
per-language LLM calls (distinct outputs): <strong>EN→FR</strong>
(<code>helix_translate-multilang-en-fr-wave4-20260616-173316.mp4</code>,
"Le soleil matinal se leva…"), <strong>EN→DE</strong>
(<code>helix_translate-multilang-en-de-wave4-20260616-173319.mp4</code>,
"Der morgendliche Sonnenaufgang…"), <strong>EN→IT</strong>
(<code>helix_translate-multilang-en-it-wave4-20260616-173419.mp4</code>,
"Il sole mattutino sorse…"); Wave 4 also re-confirmed (NOT
double-counted) RU→SR-Cyrillic
(<code>helix_translate-multilang-ru-sr-cyr-wave4-20260616-173445.mp4</code>).
<strong>§11.4.138 DEMOTION (−1):</strong> the
<code>pkg/verification</code> <strong>Multi-pass polisher</strong>
(<code>-multipass</code>) row, previously PASS via
<code>helixtranslate-cli-multipass-verify-20260615.mp4</code>, is
DEMOTED to PENDING_FORENSICS — the wave-3e regression
(<code>docs/qa/wave3e-multipass-regression/FINDING.md</code>) and wave-5
evidence prove the default-model path (<code>gpt-4</code>) fails the
<code>ValidModels</code> whitelist and silently no-ops, and even with an
explicit model the polish path can write an LLM meta-response instead of
polished prose; a tracked source fix + re-record is owed before any
re-confirmation. Net video-confirmed: 43 (Rev 14) − 1 (multipass demote)
+ 3 (Wave 4) + 4 (Wave 5) + 5 (Wave 6) = <strong>54</strong>. Wave 3f of
2026-06-16 added <strong>2 net-new</strong> — <strong>EPUB→EPUB
conversion</strong>
(<code>helix_translate-epub-output-wave3e-20260616.mp4</code>: EN .epub
→ real deepseek run → OUTPUT EPUB carries real Spanish "La Antigua
Farola la vieja farola se alzaba sola en la rocosa costa…") and
<strong>TXT→TXT translation</strong>
(<code>helix_translate-txt-translation-wave3f-20260616.mp4</code>: EN
.txt → real deepseek run → ES .txt "El río fluía tranquilamente a través
del valle verde. Los pájaros cantaban bajo la luz de la mañana."), each
ffprobe+frame-verified per §11.4.107 (790×560 yuv420p, ≥6 frames,
start-frame English-only ≠ output-frame Spanish); Wave 3f also
<strong>re-confirmed</strong> (NOT double-counted) the already-confirmed
<strong>FB2→FB2</strong>
(<code>helix_translate-fb2-convert-wave3e-20260616.mp4</code>, namespace
preserved + real Spanish), <strong>format-detection</strong>
(<code>helix_translate-format-detect-wave3e-20260616.mp4</code>,
extensionless <code>.dat</code> detected-by-content),
<strong><code>-script</code> Cyrillic↔︎Latin</strong>
(<code>helix_translate-script-conversion-wave3e-20260616.mp4</code>, RUN
A latin "Ovo je testni…" ≠ RUN B cyrillic "Систем треба…") and
<strong>markdown-translator EPUB↔︎MD round-trip</strong>
(<code>helix_translate-markdown-roundtrip-wave3f-20260616.mp4</code>, "#
El Faro Viejo / El viejo faro se alzaba solitario en la costa rocosa.")
rows with fresh frame-verified clips; running total 41 (Rev 13) + 2 =
<strong>43</strong>. Wave 3d of 2026-06-16 added <strong>4
net-new</strong>, prioritising feature-type diversity — the
<strong><code>cmd/verify-models</code> CLI</strong> (real LLMsVerifier
discovery/scoring/selection: 14 env-keyed providers with real
per-provider reachable/auth/HTTP-status/candidate/verified counts,
summary <code>total verified: 4</code>, then
<code>model-bridge list</code> selection ranking
<code>#1 novita/Sao10K/L3-8B-Stheno-v3.2 score=0.91875 … #4 siliconflow score=0.88125</code>
— <code>helix_translate-verify-models-wave3d-20260616.mp4</code>), the
<strong><code>cmd/workable-items</code> CLI</strong> (real §11.4.93
SQLite SSoT: <code>validate</code> →
<code>OK: DB … matches markdown (81 items)</code>, <code>list</code> →
real ATM-NNN rows + descriptions —
<code>helix_translate-workable-items-wave3d-20260616.mp4</code>), and
<strong>2 more providers</strong> — <strong><code>nvidia</code></strong>
("…Un anciano pescador <strong>caminaba</strong> a lo largo de la
orilla." via <code>integrate.api.nvidia.com</code>,
<code>helix_translate-provider-nvidia-translate-wave3d-20260616.mp4</code>)
and <strong><code>openrouter</code></strong> ("…Un anciano pescador
<strong>paseaba</strong> a lo largo de la orilla." via
<code>openrouter.ai/api</code>,
<code>helix_translate-provider-openrouter-translate-wave3d-20260616.mp4</code>),
the two provider outputs distinct (caminaba/paseaba) proving genuine
per-provider models, all four Wave-3d clips ffprobe+frame-verified per
§11.4.107 (790×560 yuv420p, ≥6 frames at fps=2, CLI clips show advancing
distinct sections / provider clips show start-frame English-only ≠
output-frame Spanish); the Wave-3d providers reach the
<code>llm_provider</code>/OpenAI-compat per-provider path;
<code>upstage</code> was tested but is an honest SKIP (auth=false HTTP
403), <code>together</code> SKIPped (no key) — never faked. Wave 3c of
2026-06-16 added <strong>2 net-new</strong> — the
<strong>Hyperbolic</strong> library client
(<code>pkg/translator/llm</code>) and the <strong>Fireworks</strong>
provider in the <code>llm_provider</code> submodule
(<code>llm_provider/pkg/providers/fireworks/fireworks.go</code>), each a
real-use Wave-3c <code>.mp4</code> showing a distinct EN→ES translation
through the OpenAI-compat per-provider path —
<code>helix_translate-provider-hyperbolic-translate-wave3c-20260616.mp4</code>
("Al amanecer, el sol salía sobre el tranquilo pueblo. Un anciano
pescador caminaba por la orilla…") and
<code>helix_translate-provider-fireworks-translate-wave3c-20260616.mp4</code>
("Al amanecer, el sol ascendió sobre el tranquilo pueblo. Un anciano
pescador paseaba por la orilla…"), the two outputs distinct from each
other (salía/caminaba vs ascendió/paseaba) proving genuine per-provider
models, ffprobe+frame-verified per §11.4.107 (790×560 yuv420p, ≥6 frames
at fps=2, start-frame input-only ≠ output-frame Spanish). Wave 3c also
<strong>re-confirmed</strong> (no double-count) the already-confirmed
REST <code>POST /api/v1/translate</code> and
<code>GET /health · /api/v1/version · /api/v1/providers</code> rows
against a freshly-started live <code>cmd/server</code> TLS instance
(port 9443) —
<code>helix_translate-rest-translate-wave3c-20260616.mp4</code> (real
JSON
<code>{"translated":"El sol de la mañana se alzó sobre el tranquilo pueblo.","provider":"llm-deepseek",…}</code>)
and <code>helix_translate-rest-endpoints-wave3c-20260616.mp4</code>
(real health/version/providers JSON). The
<code>GET /api/v1/verified-models</code> 404 remains an honest PENDING
gap (LLMsVerifier disabled in config — not confirmed). Wave 3b of
2026-06-16 added 3 net-new — the <strong>Gemini</strong>, <strong>Zhipu
(GLM)</strong> and <strong>Cohere</strong> library clients, each backed
by a real-use Wave-3b <code>.mp4</code> showing a distinct EN→SR-latin
translation through the per-provider path —
<code>helix_translate-provider-gemini-translate-wave3b-20260616.mp4</code>
("Staro svjetiljko stajalo je samotočno na kamenitoj obali…"),
<code>helix_translate-provider-zhipu-translate-wave3b-20260616.mp4</code>
("Stara svjetiljka stojeća samostalno na kamenitoj obali…"),
<code>helix_translate-provider-cohere-translate-wave3b-20260616.mp4</code>
(provider=cohere via
<code>-base-url https://api.cohere.ai/compatibility/v1</code>, "Stari
svjetionik stojeo je sam na kamenitoj obali… svijeća njegova vodila
brodove…"), all four outputs (incl. the re-confirmed deepseek clip
<code>helix_translate-provider-deepseek-translate-wave3b-20260616.mp4</code>
"Stari svjetionik stajao je sam na kamenitoj obali…") distinct from one
another proving genuine per-provider models, ffprobe-verified (non-zero
dur+frames, 790×560 even-dims yuv420p) + ≥6 frames extracted with one
content-verified per §11.4.107; the deepseek clip is a re-confirmation,
no double-count; Wave 3a of 2026-06-16 added 2 net-new — the
<strong>Cerebras</strong> and <strong>SambaNova</strong> per-provider
LLM clients, each backed by a real-use Wave-3a <code>.mp4</code> showing
a distinct EN→ES translation through
<code>unified-translator -provider <name></code> —
<code>helix_translate-provider-cerebras-translate-wave3a-20260616.mp4</code>
("El sol de la mañana se alzó sobre el tranquilo pueblo.") and
<code>helix_translate-provider-sambanova-translate-wave3a-20260616.mp4</code>
("El sol matutino se alzó sobre el tranquilo pueblo."), the two outputs
distinct from each other proving genuine per-provider models,
ffprobe+frame-verified per §11.4.107 with start-frame ≠ output-frame
liveness; Wave 2 of 2026-06-16 added 5 net-new — DOCX output, PDF
output, HTML→Markdown conversion,
<code>ProviderDiverseTranslators</code>, <code>EnsembleFactory</code> —
each backed by a real-use Wave-2 <code>.mp4</code> independently
validator-verified ffprobe+frame; Wave 2 also re-confirmed the
already-confirmed gRPC TranslationService row with an additional clip,
no double-count; Wave 1 added 2 — bridge-ensemble best-model selection +
bridge provider-diverse catalogue) across unified-translator (DeepSeek
translate, EPUB→TXT / HTML→EPUB / FB2→EPUB / FB2→FB2 conversions,
Serbian-Cyrillic, <code>-script</code> Cyrillic↔︎Latin, format-detection,
help/version, <code>-verify</code>, <strong><code>-multipass</code>
multi-pass polishing engine</strong>), the REST API server (POST
/api/v1/translate real DeepSeek + target_lang FIXED, TLS startup FIXED,
/health · /api/v1/version · /api/v1/providers), the gRPC server (real
EN→ES), markdown-translator (EPUB↔︎MD round-trip), preparation-translator
(FIXED real analysis), cmd/cli (real DeepSeek), monitor-server (live WS
hub), the DeepSeek library client, and the <strong>LLMsVerifier→agent
bridge CLI</strong> (<code>cmd/model-bridge</code> —
<code>helixtranslate-bridge-bestmodel-translate-20260615.mp4</code>:
in-process verify→SQLite store→select strongest
<code>novita/Sao10K/L3-8B-Stheno-v3.2</code> score 0.919 → live
<code>invoke</code> EN→ES "Good morning, friend." → "Buenos días,
amigo."; ffprobe 8.0s/80 frames, ≥6 frames content-verified) <strong>and
the LLMsVerifier→agent bridge MCP-stdio surface</strong>
(<code>cmd/model-bridge mcp</code> —
<code>helixtranslate-bridge-mcp-stdio-20260615.mp4</code>: a real
JSON-RPC MCP session over stdio —
initialize→tools/list→<code>bridge_best_model</code>→<code>bridge_invoke</code>
— where tool <code>bridge_invoke</code> returned a live verified-model
translation "Le pont relie deux rives." (isError:None); ffprobe
11.88s/10 frames, ≥6 frames content-verified). <strong>Now
video-confirmed (Wave 2):</strong> <strong>DOCX output</strong>
(Implemented; pure-Go OOXML writer
<code>pkg/ebook/docx_writer.go</code>, commit 87cd2be) — Wave-2 real-use
video <code>helix_translate-docx-output-20260616-112943.mp4</code>
(unified-translator DOCX output, openai/gpt-4, the produced
<code>.docx</code> carries real Spanish text; validator
ffprobe+frame-verified). <strong>PDF output</strong> (Implemented,
commit fb265e7; <code>pkg/ebook/pdf_writer.go</code>
Book→HTML5→<code>weasyprint</code>→Cyrillic-faithful PDF, wired into
unified-translator <code>.pdf</code> output, main.go:1160) — Wave-2
real-use video
<code>helix_translate-pdf-output-20260616-113057.mp4</code> (valid PDF
with real Spanish; validator ffprobe+frame-verified). <strong>Wave-1
bridge-translator-path blocker RESOLVED (caveat 8):</strong> the
<code>ValidModels</code>-whitelist runtime failure that blocked the
DOCX/PDF/gRPC/per-provider videos in Wave 1 is fixed — the
verified-bridge translator path now produces real per-provider
translations (Wave-2 provider clips: novita/mistral/groq/siliconflow).
<strong>Honest gaps recorded, NOT confirmed:</strong>
<code>cmd/translator</code> + <code>translate-ssh</code> +
<code>ebook-translator</code> are <strong>Obsolete (→ Fixed.md), Rev
7</strong> (<code>feature-removed</code> — the local-runtime + SSH-local
paths were removed in bridge phase-2 R-2..R-4; the default path now
sources the LLMsVerifier <code>pkg/bridge</code>; the earlier
STUB/OPERATOR-BLOCKED wording is stale and corrected);
<code>GET /api/v1/verified-models</code> returns 404 because
LLMsVerifier is disabled in config. Everything else is
<code>PENDING</code> (a recording is owed) or <code>N/A</code> (no
user-visible video applies — e.g. CLI flags, internal library types).
Absence of a runtime recording is stated honestly, never papered
over.</p>
</blockquote>
<h2 id="operator-blocked-items-read-first-11445">Operator-blocked items
(read first, §11.4.45)</h2>
<p>Two binaries are OPERATOR-BLOCKED — both require a remote SSH worker
host running llama.cpp, which is not available in this environment. The
binaries build and run as far as the missing host allows (captured in
<code>helixtranslate-cmd-blocked-binaries_20260615-172456.mp4</code>):</p>
<ul>
<li><strong><code>cmd/translate-ssh</code></strong> — SSH worker 4-step
FB2→EPUB translator. <strong>Unblock condition:</strong> a reachable SSH
host with llama.cpp + a model.</li>
<li><strong><code>cmd/ebook-translator</code></strong> — FB2
remote-translate workflow (syncs binary to remote, runs remote
translate, downloads outputs). <strong>Unblock condition:</strong> same
remote SSH/llama.cpp host.</li>
<li><strong><code>cmd/translator</code> (remote path)</strong> — remote
SSH/llama.cpp translate path is OPERATOR-BLOCKED (same host
requirement); its LOCAL path is a separate STUB ("local translation not
yet implemented").</li>
</ul>
<h2 id="coverage-summary">Coverage summary</h2>
<table>
<thead>
<tr>
<th>Metric</th>
<th>Value</th>
</tr>
</thead>
<tbody>
<tr>
<td>Headline feature total (inventory's deduplicated per-category
tally)</td>
<td><strong>417</strong></td>
</tr>
<tr>
<td>Enumerated feature rows in this doc</td>
<td><strong>497</strong> (Rev 16 = 496 Rev 14 + 1 new
<code>cmd/grpc-translate-probe</code> row — standalone gRPC
translate-round-trip probe client, Implemented + Wired +
Source-confirmed, Video-confirmation PENDING; Rev 14 = 494 Rev 13 + 2
new unified-translator rows added Wave 3f for the net-new videos:
<code>EPUB→EPUB conversion</code> + <code>TXT→TXT translation</code>;
Rev 9 = 493 Rev 8 + 1 new unified-translator
<code>HTML→Markdown conversion</code> row added that wave for the Wave-2
html-to-md video; §11.4.6 — re-derive with <code>awk</code> over
<code>^|</code> 9-column rows. The earlier "478" was NOT re-derivable
from the live table — it was a hand-tracked running tally that had
drifted; Rev 8 added the 7 missing <code>pkg/bridge</code> ensemble-seam
rows: <code>BestTranslator</code>, <code>BestTranslatorFunc</code>,
<code>EnsembleFactory</code>, <code>ProviderDiverseTranslators</code>,
<code>ProviderDiverseClients</code>, <code>ProviderDiverseModels</code>,
<code>BestClient</code>)</td>
</tr>
<tr>
<td>Implemented</td>
<td>446 (Rev 16: 445 Rev 14 + 1 new
<code>cmd/grpc-translate-probe</code> row, Implemented + Wired; Rev 14:
443 Rev 13 + 2 new unified-translator rows EPUB→EPUB conversion +
TXT→TXT translation, both Implemented + Wired + Wave-3f video-confirmed;
Rev 9: 442 Rev 8 + 1 new unified-translator HTML→Markdown conversion
row; Rev 8 was 435 Rev 7 + 7 new pkg/bridge ensemble-seam rows; was 471
Rev 6, −36 flipped to Obsolete Rev 7 — local-runtime + SSH-local
features removed in bridge phase-2 R-2..R-4)</td>
</tr>
<tr>
<td><strong>Obsolete (→ Fixed.md)</strong></td>
<td><strong>39</strong> (Rev 7 — bridge phase-2 R-2..R-4 removals:
<code>cmd/translate-ssh</code> ×9, <code>cmd/ssh-translation</code> ×2,
<code>cmd/translator</code> ×19, <code>cmd/ebook-translator</code> ×3,
<code>pkg/sshworker</code> ×2, <code>pkg/modelsbridge</code> ×1, Ollama
provider ×1, llama.cpp provider+coordinator ×2;
Reason=<code>feature-removed</code>; the KEPT distributed-API path is
NOT Obsolete)</td>
</tr>
<tr>
<td>Stub</td>
<td>4 (<code>POST /api/v1/translate/ebook</code>,
<code>POST /api/v1/preparation/analyze</code>,
<code>GET /api/v1/preparation/result/:id</code>,
<code>POST /api/v1/translate-with-verification</code>)</td>
</tr>
<tr>
<td>Not implemented (gap rows)</td>
<td>0 (PDF-write flipped gap→Implemented Rev 4, commit fb265e7; output
now EPUB/FB2/TXT/HTML/MD/DOCX/PDF)</td>
</tr>
<tr>
<td>Partial</td>
<td>2 (<code>POST /api/batch</code> on standalone
<code>pkg/api/server.go</code> — returns <code>queued</code> batch_id
only, no translation; <code>vision_engine</code> OpenCV — stub default,
real impl behind build tag) — was 3 Rev 6; <code>cmd/translator</code>
flipped to Obsolete Rev 7</td>
</tr>
<tr>
<td>Operator-blocked</td>
<td>0 (was 2 Rev 6 — <code>translate-ssh</code> +
<code>ebook-translator</code> flipped to Obsolete Rev 7: the remote
SSH/llama.cpp worker capability was removed, not merely
host-blocked)</td>
</tr>
<tr>
<td>Video-confirmed</td>
<td><strong>66</strong> (real, content-verified recordings — see
Anti-bluff note. <strong>Rev-16 / Wave 9 + Wave 8 + Wave 7 of 2026-06-16
added 12 net-new</strong> (4+5+3), of which <strong>3 are genuine
row-flips</strong> PENDING→Confirmed
(<code>POST /api/v1/translate/string</code> Wave 7, the
<code>Groq client</code> library row Wave 8,
<code>POST /api/v1/translate/batch</code> Wave 9) and 9 are
headline-counted additional distinct-scope confirmations on
already-Confirmed rows / language-pair coverage (EPUB→EPUB round-trip,
<code>-script</code> Cyrillic→Latin, preparation analysis,
<code>GET /api/v1/version</code>, preparation→translate two-stage,
EN→KO, DE→EN, FR→ES, EN→SR-latin — see Anti-bluff note for the full
per-feature mapping + the Wave-7 markdown-translator bug
found→fixed-at-4fbe581→re-record-owed + the Wave-9 honest
non-confirmable gaps). <strong>§11.4.6 cell reconciliation:</strong>
this cell had a known sync-lag behind the authoritative headline running
total (it read <strong>43</strong> at Rev 15 while the Anti-bluff-note
prose was at 54); it is corrected here to the true running total — 54
(prior authoritative) + 4 (Wave 7) + 5 (Wave 8) + 3 (Wave 9) =
<strong>66</strong>. Rev-14 / Wave 3f of 2026-06-16 added <strong>2
net-new</strong> video confirmations prioritising feature-type diversity
— <strong>EPUB→EPUB conversion</strong>
(<code>helix_translate-epub-output-wave3e-20260616.mp4</code>, EN .epub
→ real Spanish EPUB) and <strong>TXT→TXT translation</strong>
(<code>helix_translate-txt-translation-wave3f-20260616.mp4</code>, EN
.txt → real Spanish .txt), each ffprobe+frame-verified per §11.4.107;
Wave 3f also re-confirmed (NOT double-counted) FB2→FB2 /
format-detection / <code>-script</code> Cyrillic↔︎Latin / markdown
EPUB↔︎MD round-trip with fresh frame-verified clips; running total 41
(Rev 13) + 2 = <strong>43</strong>. Rev-13 / Wave 3d of 2026-06-16 added
<strong>4 net-new</strong> video confirmations prioritising feature-type
diversity — the <strong><code>cmd/verify-models</code> CLI</strong>
(<code>helix_translate-verify-models-wave3d-20260616.mp4</code>, real
LLMsVerifier discovery/scoring/selection + ranking), the
<strong><code>cmd/workable-items</code> CLI</strong>
(<code>helix_translate-workable-items-wave3d-20260616.mp4</code>, real
§11.4.93 SQLite validate+list), and <strong>2 more providers</strong>
<strong>nvidia</strong>
(<code>helix_translate-provider-nvidia-translate-wave3d-20260616.mp4</code>,
"…caminaba a lo largo de la orilla.") + <strong>openrouter</strong>
(<code>helix_translate-provider-openrouter-translate-wave3d-20260616.mp4</code>,
"…paseaba a lo largo de la orilla."), each ffprobe+frame-verified per
§11.4.107; running total 37 (Rev 12) + 4 = <strong>41</strong>.
<code>upstage</code> HTTP-403 SKIP + <code>together</code> no-key SKIP,
never faked. Rev-12 / Wave 3c of 2026-06-16 added <strong>2
net-new</strong> video confirmations — the <strong>Hyperbolic</strong>
library client
(<code>helix_translate-provider-hyperbolic-translate-wave3c-20260616.mp4</code>,
"…el sol salía…caminaba por la orilla…") and the
<strong>Fireworks</strong> <code>llm_provider</code> submodule client
(<code>helix_translate-provider-fireworks-translate-wave3c-20260616.mp4</code>,
"…el sol ascendió…paseaba por la orilla…"), each a distinct real EN→ES
translation via the OpenAI-compat per-provider path,
ffprobe+frame-verified per §11.4.107 (790×560 yuv420p, ≥6 frames at
fps=2, start-frame input-only ≠ output-frame Spanish), the two outputs
distinct proving genuine per-provider models; Wave 3c also re-confirmed
(NOT double-counted) the already-confirmed REST
<code>POST /api/v1/translate</code> +
<code>/health · /api/v1/version · /api/v1/providers</code> rows against
a live <code>cmd/server</code> TLS instance
(<code>helix_translate-rest-translate-wave3c-20260616.mp4</code> +
<code>helix_translate-rest-endpoints-wave3c-20260616.mp4</code>);
<strong>count reconciliation (§11.4.6):</strong> this cell read
<strong>32</strong> through Rev 10/11 — a sync lag behind the
Anti-bluff-note prose; it is corrected to the true running total here —
30 (Rev 9) + 2 (Wave 3a Cerebras/SambaNova) + 3 (Wave 3b
Gemini/Zhipu/Cohere) + 2 (Wave 3c Hyperbolic/Fireworks) =
<strong>37</strong>. Rev-10 / Wave 3a of 2026-06-16 added <strong>2
net-new</strong> video confirmations — the <strong>Cerebras</strong> and
<strong>SambaNova</strong> per-provider LLM clients
(<code>helix_translate-provider-cerebras-translate-wave3a-20260616.mp4</code>
+
<code>helix_translate-provider-sambanova-translate-wave3a-20260616.mp4</code>),
each a distinct real EN→ES translation via
<code>unified-translator -provider <name></code>,
ffprobe+frame-verified per §11.4.107 with start-frame ≠ output-frame
liveness, the two outputs distinct proving genuine per-provider models;
Rev-11 / Wave 3b of 2026-06-16 added <strong>3 net-new</strong> — the
<strong>Gemini</strong>, <strong>Zhipu (GLM)</strong> and
<strong>Cohere</strong> library clients
(<code>helix_translate-provider-{gemini,zhipu,cohere}-translate-wave3b-20260616.mp4</code>),
distinct EN→SR-latin per-provider translations, frame-verified per
§11.4.107; Rev-9 / Wave 2 of 2026-06-16 added <strong>5 net-new</strong>
video confirmations, each backed by a real-use Wave-2 <code>.mp4</code>
independently validator-verified ffprobe+frame §11.4.107: <strong>DOCX
output</strong>
(<code>helix_translate-docx-output-20260616-112943.mp4</code>),
<strong>PDF output</strong>
(<code>helix_translate-pdf-output-20260616-113057.mp4</code>),
<strong>HTML→Markdown conversion</strong>
(<code>helix_translate-html-to-md-20260616-114228.mp4</code>),
<strong><code>ProviderDiverseTranslators</code></strong> +
<strong><code>EnsembleFactory</code></strong> (the 4 per-provider clips
novita/mistral/groq/siliconflow —
<code>helix_translate-provider-{novita,mistral,groq,siliconflow}-translate-20260616-114*.mp4</code>).
Wave 2 also re-confirmed the already-confirmed <strong>gRPC
TranslationService</strong> row with an additional clip
<code>helix_translate-grpc-translate-20260616-113415.mp4</code> (NOT
double-counted — that row was already video-confirmed). <strong>DOCX
output and PDF output are no longer "real-artifact-only" — they are now
video-confirmed.</strong> §11.4.108 note: the Wave-1
bridge-translator-path runtime blocker (caveat 8) is RESOLVED, which is
what made the DOCX/PDF/gRPC/per-provider Wave-2 recordings genuinely
demonstrate working features (recording them in Wave 1 would have been a
§11.4.107 bluff; they were honest SKIPs then). Wave 1 added 2:
bridge-ensemble best-model selection + bridge provider-diverse
catalogue. ≈ 8.3% of 494 enumerated rows (≈ 9.8% of the 417
headline)</td>
</tr>
<tr>
<td>Video-confirmation PENDING</td>
<td>runtime/user-visible features without a recording yet</td>
</tr>
<tr>
<td>Video-confirmation N/A</td>
<td>flags, internal library types, infra middleware (no standalone
user-visible video)</td>
</tr>
</tbody>
</table>
<blockquote>
<p><strong>§11.4.6 reconciliation:</strong> The inventory's headline is
<strong>417</strong> — its own note states this is an <em>approximate,
deduplicated</em> per-category tally where "some rows are
sub-features... counted once, and a few rows span two categories
(counted once, in the first/primary category)." The inventory's
<em>detailed</em> tables actually enumerate <strong>470</strong> feature
rows. This document transcribes <strong>all 470 detailed rows
1:1</strong> so <strong>no feature is dropped</strong> (which would
itself be a bluff). The 417 headline and the 470 enumerated rows are
both reported honestly; the 53-row gap is dedup, not missing
coverage.</p>
</blockquote>
<blockquote>
<p>"Static-return" endpoints (e.g. <code>GET /api/stats</code> hardcoded
zeros, <code>GET /api/v1/status/:id</code> hardcoded
<code>completed</code>, <code>GET /api/v1/metrics</code> static,
monitor-server <code>GET /api/v1/status</code> static
<code>monitoring_active</code>) are marked <strong>Implemented</strong>
at the route level but their <code>Real-use</code> cell states the
static/hardcoded behaviour honestly so no false "working backend"
impression is created.</p>
</blockquote>
<h3 id="per-category-counts">Per-category counts</h3>
<table>
<thead>
<tr>
<th>Category</th>
<th>Count</th>
</tr>
</thead>
<tbody>
<tr>
<td>CLI</td>
<td>130</td>
</tr>
<tr>
<td>API</td>
<td>70</td>
</tr>
<tr>
<td>Web</td>
<td>23</td>
</tr>
<tr>
<td>gRPC</td>
<td>7</td>
</tr>
<tr>
<td>Library</td>
<td>81</td>
</tr>
<tr>
<td>Submodule</td>
<td>95</td>
</tr>
<tr>
<td>Infra</td>
<td>11</td>
</tr>
<tr>
<td><strong>TOTAL</strong></td>
<td><strong>417</strong></td>
</tr>
</tbody>
</table>
<hr />
<h2 id="legend">Legend</h2>
<ul>
<li><strong>Implementation:</strong> <code>Implemented</code> (real code
path present) / <code>Stub</code> (returns mock, no real work) /
<code>Partial</code> (does some but not the named work).</li>
<li><strong>Wiring:</strong> <code>Wired</code> (reachable from a binary
/ route / public API) / <code>Unwired</code> (present but not
reached).</li>
<li><strong>Real-use:</strong> what the code actually does at runtime
(honest, §11.4.6).</li>
<li><strong>Tests:</strong> <code>Covered</code>
(unit/integration/challenge present in repo) / <code>None</code> (no
test found / not asserted in inventory).</li>
<li><strong>Validation:</strong> strongest evidence available.</li>
<li><strong>Video-confirmation:</strong>
<code>Confirmed:<file></code> only with a real recording; else
<code>PENDING</code> / <code>N/A</code>.</li>
</ul>
<blockquote>
<p>Per §11.4.6, where the inventory did not read test files
line-by-line, <code>Tests</code> is recorded
<code>Not-inventoried</code> (honest unknown) rather than guessed as
Covered or None. The repo is known to carry extensive
<code>_test.go</code> suites and a <code>test/</code> tree, but
per-feature test attribution was not part of the raw inventory.</p>
</blockquote>
<hr />
<h1 id="1-cli-binaries--cmd">1. CLI Binaries — cmd/</h1>
<h2 id="cmdapi-server">cmd/api-server</h2>
<table>
<thead>
<tr>
<th>Component</th>
<th>Feature</th>
<th>Category</th>
<th>Implementation</th>
<th>Wiring</th>
<th>Real-use</th>
<th>Tests</th>
<th>Validation</th>
<th>Video-confirmation</th>
</tr>
</thead>
<tbody>
<tr>
<td>api-server</td>
<td>REST translation server</td>
<td>API/Web</td>
<td>Implemented</td>
<td>Wired</td>
<td>Gin HTTP server (:8080) proxying to gRPC backend; serves dashboard +
static files</td>
<td>Not-inventoried</td>
<td>Source-confirmed (cmd/api-server/main.go)</td>
<td>PENDING</td>
</tr>
<tr>
<td>api-server</td>
<td><code>-http-port</code> flag</td>
<td>API</td>
<td>Implemented</td>
<td>Wired</td>
<td>Override HTTP port (else <code>HTTP_PORT</code> env)</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>api-server</td>
<td><code>-http-address</code> flag</td>
<td>API</td>
<td>Implemented</td>
<td>Wired</td>
<td>Override HTTP bind address (else <code>HTTP_ADDRESS</code> env)</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>api-server</td>
<td><code>-grpc-address</code> flag</td>
<td>API</td>
<td>Implemented</td>
<td>Wired</td>
<td>Override gRPC backend address (else <code>GRPC_ADDRESS</code>
env)</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>api-server</td>
<td><code>-debug</code> flag</td>
<td>API</td>
<td>Implemented</td>
<td>Wired</td>
<td>Enable debug / Gin debug mode</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
</tbody>
</table>
<h2 id="cmdchallenge-runner">cmd/challenge-runner</h2>
<table>
<thead>
<tr>
<th>Component</th>
<th>Feature</th>
<th>Category</th>
<th>Implementation</th>
<th>Wiring</th>
<th>Real-use</th>
<th>Tests</th>
<th>Validation</th>
<th>Video-confirmation</th>
</tr>
</thead>
<tbody>
<tr>
<td>challenge-runner</td>
<td>Challenge orchestrator</td>
<td>CLI</td>
<td>Implemented</td>
<td>Wired</td>
<td>Registers + runs all anti-bluff challenges, prints PASS/FAIL/SKIP,
writes results dir</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>PENDING</td>
</tr>
<tr>
<td>challenge-runner</td>
<td><code>-parallel</code></td>
<td>CLI</td>
<td>Implemented</td>
<td>Wired</td>
<td>Run challenges in parallel</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>challenge-runner</td>
<td><code>-max-concurrency</code></td>
<td>CLI</td>
<td>Implemented</td>
<td>Wired</td>
<td>Max concurrent challenges</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>challenge-runner</td>
<td><code>-stop-on-failure</code></td>
<td>CLI</td>
<td>Implemented</td>
<td>Wired</td>
<td>Stop on first failure</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>
<tr>
<td>challenge-runner</td>
<td><code>-verbose</code></td>
<td>CLI</td>
<td>Implemented</td>
<td>Wired</td>
<td>Verbose output</td>
<td>Not-inventoried</td>
<td>Source-confirmed</td>
<td>N/A</td>
</tr>