-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathCONSTITUTION.html
More file actions
636 lines (636 loc) · 28.1 KB
/
Copy pathCONSTITUTION.html
File metadata and controls
636 lines (636 loc) · 28.1 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
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8" />
<meta name="generator" content="pandoc" />
<meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes" />
<title>HelixTranslate Constitution</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: min(4vw, 1.5em);}
div.column{flex: auto; 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">HelixTranslate Constitution</h1>
</header>
<h1 id="helixtranslate-constitution">HelixTranslate — Constitution</h1>
<blockquote>
<p><strong>Status:</strong> Active. This document is the project’s
authoritative rule set. When a rule here conflicts with
<code>CLAUDE.md</code>, <code>AGENTS.md</code>, or any guide, the
Constitution wins.</p>
</blockquote>
<h2 id="inherited-from-constitutionconstitution.md">INHERITED FROM
constitution/Constitution.md</h2>
<p>This constitution <strong>extends</strong> the Helix Universal
Constitution at <code>constitution/Constitution.md</code>. All clauses
there apply unless explicitly overridden below with an explicit
<code>Override §X.Y</code> section. The universal clauses are
authoritative for any topic this file does not cover; the
project-specific rules below (CONST-033 … CONST-036 and the Definition
of Done) extend them and never weaken any universal clause. When this
file disagrees with the constitution submodule, the constitution
wins.</p>
<h2 id="mission">Mission</h2>
<p>See README.md.</p>
<h2 id="mandatory-standards">Mandatory Standards</h2>
<ol type="1">
<li><strong>Reproducibility:</strong> every change is reproducible from
a clean clone
(<code>git clone <repo> && <project bootstrap></code>);
no hidden steps.</li>
<li><strong>Tests track behavior, not code:</strong> test what the
user-visible behavior is, not what the implementation looks like.</li>
<li><strong>No silent skips, no silent mocks above unit
tests.</strong></li>
<li><strong>Conventional Commits</strong> for all commits.</li>
<li><strong>SSH-only for git operations</strong> (<code>git@…</code>);
HTTPS prohibited.</li>
</ol>
<h2 id="numbered-rules">Numbered Rules</h2>
<!-- Rules are numbered CONST-NNN. New rules append. Removed rules
keep their number with a "**Retired:** …" line. -->
<!-- BEGIN host-power-management addendum (CONST-033) -->
<h3 id="const-033-host-power-management-is-forbidden">CONST-033 — Host
Power Management is Forbidden</h3>
<p><strong>Status:</strong> Mandatory. Non-negotiable. Applies to every
project, submodule, container entry point, build script, test,
challenge, and systemd unit shipped from this repository.</p>
<p><strong>Rule:</strong> No code in this repository may invoke a
host-level power- state transition (suspend, hibernate, hybrid-sleep,
suspend-then- hibernate, poweroff, halt, reboot, kexec) on the host
machine. This includes — but is not limited to:</p>
<ul>
<li><code>systemctl {suspend,hibernate,hybrid-sleep,suspend-then-hibernate,poweroff,halt,reboot,kexec}</code></li>
<li><code>loginctl {suspend,hibernate,hybrid-sleep,suspend-then-hibernate,poweroff,halt,reboot}</code></li>
<li><code>pm-{suspend,hibernate,suspend-hybrid}</code></li>
<li><code>shutdown {-h,-r,-P,-H,now,--halt,--poweroff,--reboot}</code></li>
<li>DBus calls to
<code>org.freedesktop.login1.Manager.{Suspend,Hibernate,HybridSleep,SuspendThenHibernate,PowerOff,Reboot}</code></li>
<li>DBus calls to
<code>org.freedesktop.UPower.{Suspend,Hibernate,HybridSleep}</code></li>
<li><code>gsettings set ... sleep-inactive-{ac,battery}-type</code> to
any value other than <code>'nothing'</code> or <code>'blank'</code></li>
</ul>
<p><strong>Why:</strong> The host runs mission-critical parallel
CLI-agent and container workloads. On 2026-04-26 18:23:43 the host was
auto- suspended by the GDM greeter’s idle policy mid-session, killing
HelixAgent and 41 dependent services. Recurring memory-pressure SIGKILLs
of <code>user@1000.service</code> (perceived as “logged out”) have the
same outcome. Auto-suspend, hibernate, and any power-state transition
are unsafe for this host.</p>
<p><strong>Defence in depth (mandatory artifacts in every
project):</strong> 1.
<code>scripts/host-power-management/install-host-suspend-guard.sh</code>
— privileged installer, manual prereq, run once per host with sudo.
Masks <code>sleep.target</code>, <code>suspend.target</code>,
<code>hibernate.target</code>, <code>hybrid-sleep.target</code>; writes
<code>AllowSuspend=no</code> drop-in; sets logind
<code>IdleAction=ignore</code> and <code>HandleLidSwitch=ignore</code>.
2.
<code>scripts/host-power-management/user_session_no_suspend_bootstrap.sh</code>
— per-user, no-sudo defensive layer. Idempotent. Safe to source from
<code>start.sh</code> / <code>setup.sh</code> /
<code>bootstrap.sh</code>. 3.
<code>scripts/host-power-management/check-no-suspend-calls.sh</code> —
static scanner. Exits non-zero on any forbidden invocation. 4.
<code>pkg/challenge_runner/scripts/host_no_auto_suspend_challenge.sh</code>
— asserts the running host’s state matches layer-1 masking. 5.
<code>pkg/challenge_runner/scripts/no_suspend_calls_challenge.sh</code>
— wraps the scanner as a challenge that runs in CI /
<code>run_all_challenges.sh</code>.</p>
<p><strong>Enforcement:</strong> Every project’s CI /
<code>run_all_challenges.sh</code> equivalent MUST run both challenges
(host state + source tree). A violation in either channel blocks merge.
Adding files to the scanner’s <code>EXCLUDE_PATHS</code> requires an
explicit justification comment identifying the non-host context.</p>
<p><strong>See also:</strong> <code>docs/HOST_POWER_MANAGEMENT.md</code>
for full background and runbook.</p>
<!-- END host-power-management addendum (CONST-033) -->
<!-- BEGIN llmsverifier-single-source-of-truth addendum (CONST-034) -->
<h3
id="const-034-llmsverifier-is-the-single-source-of-truth-for-all-llm-models">CONST-034
— LLMsVerifier is the Single Source of Truth for All LLM Models</h3>
<p><strong>Status:</strong> Mandatory. Non-negotiable. Applies to all
translation workflows, API endpoints, CLI commands, distributed workers,
and WebSocket sessions.</p>
<p><strong>Rule:</strong> LLMsVerifier SHALL be the EXCLUSIVE and ONLY
source of truth for all LLM models available to HelixTranslate. No model
may be used in HelixTranslate that does not originate from LLMsVerifier
and pass its verification gate.</p>
<p><strong>Requirements:</strong> 1. <strong>F-SSOT-001:</strong> Only
models verified by LLMsVerifier are eligible for use. 2.
<strong>F-SSOT-002:</strong> HelixTranslate MUST NOT maintain any
independent model registry, hardcoded model lists, or fallback model
definitions outside of LLMsVerifier-provided data. 3.
<strong>F-SSOT-003:</strong> All model metadata (name, ID, capabilities,
token limits, pricing, features) SHALL be retrieved from LLMsVerifier at
runtime or build time. 4. <strong>F-SSOT-004:</strong> Model information
SHALL be cached locally with configurable TTL, but cache invalidation
MUST trigger re-fetch from LLMsVerifier. 5. <strong>F-SSOT-005:</strong>
If LLMsVerifier is unreachable, HelixTranslate SHALL either use expired
cache with warning or enter degraded mode with explicit user
notification. No unauthorized models SHALL be used. 6.
<strong>F-GATE-001:</strong> ONLY models that have PASSED LLMsVerifier
validation, verification, and scoring pipeline SHALL be presented to end
users. 7. <strong>F-GATE-002:</strong> Models MUST satisfy ALL three
criteria: (a) validation passed, (b) verification passed, (c) positive
scoring (> 0 overall score). 8. <strong>F-GATE-003:</strong> The
scoring algorithm:
<code>Overall Score = (Responsiveness x 0.30) + (Code Capability x 0.25) + (Feature Richness x 0.25) + (Reliability x 0.20)</code>.
Models with score <= 0 SHALL be rejected. 9.
<strong>F-GATE-004:</strong> Models MUST have
<code>VerificationStatus == "verified"</code> AND
<code>CanSeeCode == true</code> to be eligible. 10.
<strong>F-GATE-005:</strong> Models MUST have
<code>AffirmativeResponse == true</code> to be eligible.</p>
<p><strong>Enforcement:</strong> Any code change that introduces a
hardcoded model list, bypasses the verification gate, or uses an
unverified model constitutes a Constitution violation and MUST be
reverted.</p>
<!-- END llmsverifier-single-source-of-truth addendum (CONST-034) -->
<!-- BEGIN anti-bluff-testing addendum (CONST-035) -->
<h3 id="const-035-anti-bluff-testing-constitution-mandatory">CONST-035 —
Anti-Bluff Testing Constitution (MANDATORY)</h3>
<p><strong>Status:</strong> Mandatory. Non-negotiable. Applies to every
test, challenge, HelixQA bank entry, and CI pipeline in this repository
and all submodules.</p>
<p><strong>§35.1 The Problem (Historical Mandate)</strong></p>
<blockquote>
<p>“We had been in position that all tests do execute with success and
all Challenges as well, but in reality the most of the features does not
work and can’t be used!”</p>
</blockquote>
<p>This is the <strong>worst possible outcome</strong>: green tests +
broken features. Every rule below exists to prevent this.</p>
<p><strong>§35.2 The Six Anti-Bluff Rules</strong></p>
<p>Every test, Challenge, and HelixQA bank entry MUST:</p>
<ol type="1">
<li><p><strong>Assert on a concrete end-user-visible outcome</strong> —
translated text content, DB row, downloadable file, visible dashboard
element, API response body. NOT just “no error” or “200 OK” or “exit
code 0”.</p></li>
<li><p><strong>Run against the real system</strong> — mocks ONLY in unit
tests (<code>go test -short</code>). ALL other test types MUST use real
services, real databases, real LLM providers (or documented
<code>SKIP-OK: #<ticket></code>).</p></li>
<li><p><strong>Include a matching negative assertion</strong> — test
MUST fail when the feature is broken. E.g., if testing “English→Serbian
translation”, also assert the output is NOT in English.</p></li>
<li><p><strong>Emit copy-pasteable evidence</strong> — response body
snippet, screenshot filename, DB row dump, log excerpt, video
timestamp.</p></li>
<li><p><strong>Verify “fails when feature is removed”</strong> —
deliberately break the feature (comment out implementation, change API
key to invalid), re-run test, test MUST FAIL.</p></li>
<li><p><strong>No blind shells</strong> — no
<code>&& echo PASS</code>, no <code>|| true</code>, no
<code>tee</code> exit-code laundering, no
<code>test -f file && echo "PASS"</code> without checking file
content.</p></li>
</ol>
<p><strong>§35.3 HelixQA-Specific Anti-Bluff Rules</strong></p>
<ul>
<li>Bank entries declare <strong>executable actions</strong> (never
prose).</li>
<li>Each entry declares <strong>concrete success predicates</strong>:
<code>assertBodyContains: 'translated text'</code>,
<code>assertVisible: 'Translation Complete'</code>.</li>
<li><strong>Stagnation guard</strong> — frame N+1 identical to N for
>10 seconds = FAIL.</li>
<li>Vision-model <code>verified=true</code> with empty/tautological
reasoning = <code>INCONCLUSIVE</code> (not PASS).</li>
<li><code>IsBlankScreenshot()</code> must pass before any vision
analysis.</li>
</ul>
<p><strong>§35.4 Functional Probe Floor</strong></p>
<ul>
<li>TCP-open is the FLOOR, not the ceiling.</li>
<li>PostgreSQL → <code>SELECT 1</code> returns <code>1</code>.</li>
<li>Redis → <code>PING</code> returns <code>PONG</code>.</li>
<li>API → <code>GET /health</code> returns
<code>{"status": "healthy"}</code> with non-empty body.</li>
<li>Translation → actual translated text in response, not just
session_id.</li>
</ul>
<p><strong>§35.5 Evidence Requirements</strong></p>
<ul>
<li>Every PASS must carry positive evidence captured during
execution.</li>
<li>No metadata-only PASS, no configuration-only PASS, no
“absence-of-error” PASS.</li>
<li>Evidence types: API response bodies, downloaded files, database
query results, browser console output, screenshots, video frames.</li>
</ul>
<p><strong>§35.6 Bluff Taxonomy (FORBIDDEN Patterns)</strong></p>
<table>
<colgroup>
<col style="width: 31%" />
<col style="width: 25%" />
<col style="width: 42%" />
</colgroup>
<thead>
<tr>
<th>Bluff Type</th>
<th>Example</th>
<th>Why It’s Wrong</th>
</tr>
</thead>
<tbody>
<tr>
<td><strong>Wrapper bluff</strong></td>
<td>Test asserts function returns, but caller ignores return value</td>
<td>Passes but feature unused</td>
</tr>
<tr>
<td><strong>Contract bluff</strong></td>
<td>System advertises capability but rejects it in dispatch</td>
<td>Advertises what it can’t do</td>
</tr>
<tr>
<td><strong>Structural bluff</strong></td>
<td>Checks file exists but doesn’t verify content</td>
<td>File exists but is empty/corrupt</td>
</tr>
<tr>
<td><strong>Comment bluff</strong></td>
<td>Code comment promises behavior code doesn’t have</td>
<td>Tests comment, not code</td>
</tr>
<tr>
<td><strong>Skip bluff</strong></td>
<td><code>t.Skip("not running yet")</code> without <code>SKIP-OK</code>
marker</td>
<td>Hides broken test</td>
</tr>
</tbody>
</table>
<p><strong>§35.7 Mutation Testing (Mandatory)</strong></p>
<ul>
<li>Every challenge MUST have a paired mutation test.</li>
<li>Mutation deliberately breaks the feature → the challenge MUST then
FAIL.</li>
<li>A challenge without a paired mutation = BLUFF challenge =
Constitution violation.</li>
</ul>
<p><strong>§35.8 Audit Ritual</strong></p>
<p>Every Full-QA cycle MUST: 1. Pick 5 random tests + 5 random
challenges. 2. Comment out the target implementation. 3. Re-run
tests/challenges. 4. Confirm they FAIL. 5. Restore implementation. 6.
Document results in session report.</p>
<p><strong>§35.9 User Mandate (NON-NEGOTIABLE)</strong></p>
<p>The bar is NOT “tests pass” but <strong>“users can use the
feature.”</strong> A translation that “completes” but produces garbage
text is a FAIL. An API that returns 200 but with wrong content is a
FAIL. A dashboard that “loads” but shows no data is a FAIL.</p>
<!-- END anti-bluff-testing addendum (CONST-035) -->
<!-- BEGIN helixqa-mandate addendum (CONST-036) -->
<h3 id="const-036-helixqa-is-the-sole-authorized-qa-tool">CONST-036 —
HelixQA is the Sole Authorized QA Tool</h3>
<p><strong>Status:</strong> Mandatory. Non-negotiable.</p>
<p><strong>Rule:</strong> All automated UI/UX testing of the Web
Dashboard and API endpoints MUST use HelixQA. No custom Playwright
scripts, no curl-based test harnesses outside HelixQA banks.</p>
<p><strong>Requirements:</strong> 1. <strong>HelixQA-only for Web
Dashboard and API testing.</strong> 2. <strong>Vision-driven
only.</strong> Screenshot → LLM analysis → action decision. No hardcoded
selectors, no sleep timers. 3. <strong>Universal Solution
Principle.</strong> When HelixQA cannot interact with a HelixTranslate
UI element, the fix MUST be implemented in HelixQA itself, never by
adding test hooks to HelixTranslate. 4. <strong>Live log
monitoring.</strong> Every session streams API logs, gRPC logs,
translation logs. 5. <strong>Screen-state tracking.</strong> Frame N vs
N+1. Stagnation >10s = critical failure. 6. <strong>Executable
actions in banks</strong>, never prose. 7. <strong>Video mandatory for
Web Dashboard sessions.</strong> Screenshots at every step. 8.
<strong>Evidence validation.</strong> Post-translation must contain
actual translated text, not placeholder.</p>
<!-- END helixqa-mandate addendum (CONST-036) -->
<h2 id="definition-of-done">Definition of Done</h2>
<p>A change is done when:</p>
<ol type="1">
<li>The code change is committed.</li>
<li>All project-level tests pass on a clean clone.</li>
<li>All challenges in <code>pkg/challenge_runner/scripts/</code> pass on
the running host.</li>
<li>Governance docs (<code>CONSTITUTION.md</code>,
<code>AGENTS.md</code>, <code>CLAUDE.md</code>) are coherent with the
change.</li>
<li>All HelixQA banks pass for all configured platforms.</li>
<li>Anti-bluff audit (CONST-035 §35.8) passes — random tests confirmed
to fail when feature removed.</li>
<li>Every modified/added feature has a registered challenge.</li>
<li>Evidence from the most recent QA session is archived in
<code>docs/reports/qa-sessions/</code>.</li>
</ol>
<h2 id="see-also">See also</h2>
<ul>
<li><code>README.md</code> — project overview, quickstart.</li>
<li><code>AGENTS.md</code> — guidance for AI coding agents (Codex,
Cursor, etc.).</li>
<li><code>CLAUDE.md</code> — guidance specifically for Claude Code.</li>
<li><code>docs/HOST_POWER_MANAGEMENT.md</code> — CONST-033 background
and runbook.</li>
<li><code>docs/research/llms_verifer/</code> — LLMsVerifier integration
materials.</li>
<li><code>docs/research/helix_qa/</code> — HelixQA integration
materials.</li>
</ul>
<h2 id="const-035-anti-bluff-operative-rule-mandatory">CONST-035 —
Anti-Bluff Operative Rule (MANDATORY)</h2>
<blockquote>
<p>“We had been in position that all tests do execute with success and
all Challenges as well, but in reality the most of the features does not
work and can’t be used! This MUST NOT be the case.”</p>
</blockquote>
<p><strong>The operative rule:</strong> Execution of tests and
Challenges MUST guarantee the quality, the completion and full usability
by end users of the product.</p>
<ul>
<li>A green test or challenge for a feature that does not actually work
is a <strong>BLUFF</strong> and is <strong>FORBIDDEN</strong>.</li>
<li>Every test must assert concrete user-visible outcomes, not just
internal state.</li>
<li>Every challenge must run real code and verify real behavior;
grep/file-existence checks are <strong>NOT sufficient</strong>.</li>
<li>Mutation testing is <strong>MANDATORY</strong>: deliberately break
the feature → the test/challenge <strong>MUST then FAIL</strong>.</li>
<li>The bar for shipping is <strong>NOT “tests pass”</strong> but
<strong>“users can use the feature.”</strong></li>
<li>No false-success results are tolerable.</li>
</ul>
<h2 id="mirrored-constitution-anchors-11.4.15411.4.157">Mirrored
constitution anchors (§11.4.154–§11.4.157)</h2>
<p>Condensed consumer-side mirrors of the four universal anchors most
recently added to the constitution submodule. Canonical text lives in
<code>constitution/Constitution.md</code>; these mirrors carry the
literal <code>11.4.N</code> tokens the propagation gates
(<code>CM-COVENANT-114-154/155/156/157-PROPAGATION</code>) require
(§11.4.35). Faithful condensations — no clause weakened, nothing
invented.</p>
<p><strong>§11.4.154 — Window-scoped capture + fresh-corpus rotation for
feature/QA recordings (User mandate, 2026-06-15).</strong> Refines
§11.4.2/.5/.107/.153 recording discipline with two capture-hygiene
invariants. <strong>(A) Window-scoped, NOT whole-screen</strong> — every
feature/QA video MUST capture ONLY the window/surface of the app/service
under test (GUI window / CLI-TUI terminal pane / web tab-viewport /
device-emulator frame), NEVER the whole desktop/monitor or unrelated
windows (whole-desktop capture leaks operator-private content
§11.4.10/.83, dilutes the §11.4.107 liveness/freeze oracle, breaks the
§11.4.137 OCR/ROI oracle); target by stable identity (window id/title,
device serial, browser context, tmux target) per §11.4.111, never a
fixed full-screen index; platform genuinely cannot capture below
whole-screen ⇒ honest §11.4.3 SKIP + tracked migration item. <strong>(B)
Fresh-corpus rotation</strong> — when a new recording run for a scope
begins, the agent’s OWN prior in-scope stale recordings at the raw
recording path MUST be removed FIRST so the live corpus reflects the
current run (§11.4.107 not-stale + §11.4.86 roster-freshness); “remove
old” = the agent’s own prior recordings for the SAME scope/project ONLY,
NEVER another project’s/operator-authored files (uncertain ⇒ surface,
don’t delete §11.4.122/§9.2); committed
<code>docs/qa/<run-id>/</code> evidence (§11.4.83) is the durable
record, NOT rotated. Classification: universal (§11.4.17). Composes
§11.4.2/.5/.10/.83/.86/.107/.111/.122/.128/.137/.153/§9.2/.6. Canonical
authority: constitution submodule <code>Constitution.md</code>
§11.4.154. Non-compliance is a release blocker.</p>
<p><strong>§11.4.155 — Project-name-prefixed feature/QA recording
filenames (User mandate, 2026-06-15).</strong> Every recorded video the
project produces (§11.4.153 real-use, §11.4.154 window-scoped, §11.4.128
always-on device, any raw/curated artefact at the recording path + the
committed <code>docs/qa/<run-id>/</code> trail §11.4.83) MUST have
a filename that STARTS WITH the PROJECT-NAME prefix, ALWAYS; an
unprefixed recording is a §11.4.155 violation (un-greppable +
un-attributable multi-project corpus — the §11.4.151 identify-and-grep
failure on the recording axis). <strong>Prefix resolution (closed-set,
deterministic — §11.4.6, IDENTICAL to §11.4.151):</strong> (1)
<code>HELIX_RELEASE_PREFIX</code> from <code>.env</code> (git-ignored
§11.4.30, documented in tracked <code>.env.example</code> §11.4.77) else
(2) lowercased snake_case project-root dir name §11.4.29; SAME prefix
for EVERY recording in a checkout; canonical form
<code><PREFIX>---<feature-or-scope>---<run-id>.<ext></code>;
MUST equal the §11.4.151-resolved release-tag prefix (divergence is
itself a violation — one project, one name). Honest boundary (§11.4.6):
the prefix guarantees attribution + greppability, NOT content validity
(still §11.4.107/.137/.153) and does NOT relax §11.4.154’s
window-scope/rotation. Classification: universal (§11.4.17). Composes
§11.4.151/.128/.153/.154/.111/.83/.6/.29/.30/.35/.77/.86. Canonical
authority: constitution submodule <code>Constitution.md</code>
§11.4.155. Non-compliance is a release blocker.</p>
<p><strong>§11.4.156 — All CI/CD automation (GitHub Actions / GitLab
pipelines / equivalents) MUST be disabled (User mandate,
2026-06-15).</strong> Every repository this Constitution governs — main
repo, this constitution submodule, every owned + nested submodule we
author and push — MUST ship with ALL server-side CI/CD automation
DISABLED: no push to any owned upstream may trigger a GitHub Actions
run, GitLab pipeline, or equivalent
(Jenkins/CircleCI/Travis/Drone/Woodpecker/Bitbucket/Azure, any
<code>on: push</code>/<code>schedule</code>/<code>workflow_dispatch</code>).
GENERALISES + makes ABSOLUTE the §11.4.75 Layer-5 posture across ALL
governed repos; enforcement migrates to the LOCAL §11.4.75 git-hook
ritual + §11.4.40 pre-tag sweep, never a remote runner. ALL hold:
<strong>(A)</strong> zero active root-level
<code>.github/workflows/*.yml|yaml</code> / <code>.gitlab-ci.yml</code>
/ equivalent; <strong>(B)</strong> “disabled” = a push triggers ZERO
runs — delete OR rename to a non-trigger name (the §11.4.75
<code>.disabled-local-only</code> pattern); a
live-<code>on:</code>+<code>if:false</code> workflow is NOT compliant;
<strong>(C)</strong> scope = repos we author+push — vendored/third-party
nested configs are INERT, OUT of scope (§11.4.29 vendor-exempt);
<strong>(D)</strong> no new CI may be added; <strong>(E)</strong>
pre-push verify the tracked workflow set is empty for authored repos.
Honest boundary (§11.4.6): file-level disabling stops FILE-triggered
runs, NOT provider-side server settings (org-default required workflows,
branch-protection checks) — the operator turns those off. Composes
§11.4.75/.29/.6/.40/.42/.109/.113/§2.1. Classification: universal
(§11.4.17). The parent repo currently has NO active CI workflow files —
already §11.4.156-compliant. Canonical authority: constitution submodule
<code>Constitution.md</code> §11.4.156. Non-compliance is a release
blocker.</p>
<p><strong>§11.4.157 — GEMINI.md maintained in lockstep with CLAUDE.md /
AGENTS.md / QWEN.md (User mandate, 2026-06-15).</strong>
<code>GEMINI.md</code> is a FIRST-CLASS governance context carrier EQUAL
to <code>CLAUDE.md</code>/<code>AGENTS.md</code>/<code>QWEN.md</code>,
never optional/best-effort. ALL hold: <strong>(A)</strong> five-carrier
lockstep — no governance change is complete until <code>GEMINI.md</code>
carries it alongside the other three mirrors (added to the §11.4.26
propagation + cross-reference set explicitly); <strong>(B)</strong> no
silent drift — <code>GEMINI.md</code> lagging the other mirrors’ highest
rule is a §11.4.157 violation (§11.4.65-class), back-fill required;
<strong>(C)</strong> equal status — <code>GEMINI.md</code> restates the
SAME literal <code>11.4.N</code> anchors the propagation gates require
(§11.4.35), fleet count INCLUDES <code>GEMINI.md</code>;
<strong>(D)</strong> consumer projects’ own CLAUDE/AGENTS/QWEN/GEMINI
bind too (§11.4.35). Honest boundary (§11.4.6): claiming
<code>GEMINI.md</code> “in sync” while a back-fill is incomplete is
itself a §11.4.157 violation. Composes
§11.4.26/.35/.17/.44/.65/.140/.156. Classification: universal
(§11.4.17). This parent repo now maintains <code>CLAUDE.md</code> +
<code>AGENTS.md</code> + <code>QWEN.md</code> + <code>GEMINI.md</code>
in lockstep alongside <code>CONSTITUTION.md</code>; every future
governance edit MUST land in all carriers. Canonical authority:
constitution submodule <code>Constitution.md</code> §11.4.157.
Non-compliance is a release blocker.</p>
</body>
</html>