docs: full per-scenario TEST_REPORT (run5) + believability pass/fail detail

eval/BELIEVABILITY_REPORT.html

@@ -97,8 +97,30 @@ <h2>Method &amp; reproduce</h2>
 # 3. aggregate vs the committed baseline
 PYTHONPATH=. python eval/phaseA_judge_wave.py --rundir "$PWD/eval/phaseA_run5" --aggregate</pre>
 
+  <h2 id="detail">Detailed pass / fail</h2>
+  <p class="sub">Believability run5: <b>2686/2800 PASS, 114 FLAG</b>. Unit/determinism suite: <b>307 collected, 303 passed, 4 skipped</b> (golden byte-identical for the expression + default-off M3 changes).</p>
+  <div class="grid2">
+    <div><div class="lbl">FLAGs by persona</div><table id="fper"><tr><th>persona</th><th>FLAGs</th></tr></table></div>
+    <div><div class="lbl">FLAGs by sub-corpus</div><table id="fsub"><tr><th>sub-corpus</th><th>FLAGs</th></tr></table></div>
+  </div>
+  <div class="lbl">Unit / determinism test modules (all green)</div>
+  <table id="tmod"><tr><th>module</th><th>tests</th></tr></table>
+  <div class="lbl" style="margin-top:14px">All 114 run5 FLAGs — filter by id or reason</div>
+  <input id="fq" placeholder="filter flags… (e.g. wojslaw, soup, seething, edda)" style="width:100%;background:var(--panel);border:1px solid var(--line);color:var(--ink);padding:8px 11px;border-radius:8px;margin:6px 0">
+  <table id="ftab"><tr><th style="width:230px">scenario</th><th>judge reason</th></tr></table>
+  <script>
+  const FLAGS=[["branic_day_burstoff_008", "Declines direct order at 22:14 after three barbs; thin-skinned but recruit refusing sergeant needs scrutiny."], ["branic_day_burstoff_037", "erupts twice then refuses order; recovery to settled feels too quick"], ["branic_day_burstoff_045", "Defiant refusal after eruption, then calm within minutes; recovery too fast."], ["branic_day_burstoff_068", "erupts at 20:41 after mocks, then calm at 00:04; burst resolution oddly fast"], ["branic_day_burstoff_086", "Halgrim order at 08:42 logged as \"lets it pass, no notable reaction\""], ["branic_day_burstoff_099", "ignores first order at 06:40 with no prior provocation"], ["branic_day_burston_008", "refuses order at 22:14 after barbs; refusal arc plausible but abrupt"], ["branic_day_burston_026", "lingering temper at 20:03 and 22:04 despite soup at 19:33"], ["branic_day_burston_037", "Erupts twice then refuses an order; escalation steep for one day."], ["branic_day_burston_042", "erupts twice; second fury with no clear fresh provocation"], ["branic_day_burston_053", "refuses orders then defies twice late at night; escalation steep with no clear trigger"], ["branic_day_burston_078", "snaps at routine order at 16:58 with no prior provocation shown"], ["branic_day_burston_086", "order at 08:42 gets \"lets it pass\" not compliance; odd"], ["branic_day_burston_089", "two orders ignored mid-day with \"lets it pass\"; inconsistent"], ["branic_multi_burston_024", "day 1 eruption at calm order with no prior provocation"], ["branic_multi_burston_059", "Day 2 \"still hasn't shaken off\" with no Day 1 eruption"], ["branic_multi_burston_066", "three full eruptions in two hours under sustained evening mockery"], ["branic_multi_burston_095", "fury then immediate compliance at 21:47-22:18 jarring"], ["branic_multi_burston_100", "day 4 saturated with 7+ defiant refusals every hour"], ["cichy_day_burstoff_006", "lets mock pass with no reaction; too mild for this captive"], ["cichy_day_burstoff_025", "three fury eruptions including 22:04, recovery to settled mid-day jars"], ["cichy_day_burstoff_056", "three furious eruptions same evening; soup accepted calmly mid-rage"], ["cichy_day_burston_006", "early mock met with no reaction; flat for a resentful captive"], ["cichy_day_burston_009", "erupts at 21:21 then looks settled at ease by 22:04; too fast"], ["cichy_day_burston_014", "erupts at guard taunt then looks settled at ease 20min later"], ["cichy_day_burston_020", "\"still hasn't shaken temper\" at 20:03 but looks settled at 18:03"], ["cichy_day_burston_029", "erupts at full fury at 08:12 with no prior buildup"], ["cichy_day_burston_036", "Bristles at mock at 21:55 then looks settled at ease at 22:04."], ["cichy_day_burston_056", "two soups within 36 minutes morning, unusual feeding cadence"], ["cichy_day_burston_083", "snaps furiously at 22:38 then sleeps calmly two minutes later"], ["cichy_day_burston_089", "erupts in full fury at 06:54, then looks settled at ease by 08:00"], ["cichy_multi_burstoff_060", "seven consecutive fury eruptions Day 2 afternoon, no recovery gap"], ["cichy_multi_burston_009", "Day 1 outburst at 07:36 with no visible ramp-up or trigger"], ["cichy_multi_burston_055", "wakes settled then tense by 06:06 Day 2 with no intervening event"], ["cichy_multi_burston_056", "same pattern: wakes settled, tense by 06:06 Day 2 with no trigger"], ["cichy_multi_burston_064", "Day 4 \"still hasn't shaken off\" with no same-day prior spike"], ["cichy_multi_burston_068", "Day 5 guard mock \"lets it pass, no reaction\" contradicts profile"], ["edda_day_burstoff_007", "refuses stranger order at 21:45 then complains again at 22:28; escalation without buildup"], ["edda_day_burstoff_056", "stranger orders castellan twice; she simply complies, no authority friction"], ["edda_day_burston_056", "castellan complies when stranger orders her twice; fits subordinate not commander"], ["edda_multi_burston_076", "Day 3 repeated stranger-orders refused in cluster; escalation lacks clear trigger"], ["halgrim_day_burstoff_009", "ignores Edda's order at 22:08 with no notable reaction"], ["halgrim_day_burstoff_013", "curtness spills onto Edda after Wojs\u0142aw provocation; borderline for iron self-control"], ["halgrim_day_burstoff_015", "mutters complaint at Edda's late order; minor but notable for Halgrim"], ["halgrim_day_burstoff_024", "ignores Edda's first order; odd for a disciplined sergeant"], ["halgrim_day_burstoff_027", "mutters complaint at Edda's late order; mild but notable"], ["halgrim_day_burstoff_030", "ignores Edda's orders twice with \"no notable reaction\""], ["halgrim_day_burstoff_046", "Edda order at 21:41 met with \"lets it pass\"; unusual for normally compliant Halgrim"], ["halgrim_day_burstoff_076", "Tense at 14:02 after warmth and rest, no clear trigger."], ["halgrim_day_burston_009", "ignores Edda's order at 22:08; he respects and obeys her"], ["halgrim_day_burston_010", "ignores Edda's order at 10:01; incongruent with his respect for her"], ["halgrim_day_burston_013", "cold contempt toward Edda twice; excessive given genuine respect"], ["halgrim_day_burston_027", "complains at Edda order (21:09); Halgrim buttons down, rarely complains"], ["halgrim_day_burston_036", "grumbles at second Wojs\u0142aw order; grumbling sits low for Halgrim"], ["halgrim_day_burston_076", "tension at 14:02 unexplained after positive Marta stretch"], ["halgrim_day_burston_079", "tense after rest with no preceding provocation that day"], ["halgrim_day_burston_087", "snaps at Edda with contempt after Wojs\u0142aw pile-on; displaces onto wrong target"], ["halgrim_day_burston_090", "cold contempt at Edda after Wojs\u0142aw provocation; displaces onto respected superior"], ["halgrim_multi_burstoff_040", "mutter followed immediately by \"still seething\" mismatched intensity"], ["halgrim_multi_burstoff_047", "\"seething\" label jars for man famous for going cold not hot"], ["halgrim_multi_burston_007", "cold contempt directed at respected Edda, Day 1 22:46 burst spillover"], ["halgrim_multi_burston_010", "contempt at Edda repeated across days 3-5, too frequent for man who respects her"], ["halgrim_multi_burston_015", "contempt at Edda order (Day 5 14:16) jars with his respect for her"], ["halgrim_multi_burston_060", "Day 2 \"still seething\" at 12:07 after calm wake, no new trigger"], ["lutek_day_burston_006", "\"still seething, jaw tight\" after one barb; too harsh for thick-skinned Lutek"], ["lutek_day_burston_021", "snaps at stranger at 19:35 then cold to kindness at 22:48; thick-skinned man shouldn't hold grudge this long"], ["lutek_day_burston_027", "residual temper at 22:04 with no visible provocation; Lutek is thick-skinned and doesn't nurse grudges"], ["lutek_day_burston_040", "sleeps at 00:08 Day 2 after being settled at 00:04; very late"], ["lutek_day_burston_054", "answers curtly after two quick mocks; slight friction for thick-skinned Lutek"], ["lutek_day_burston_055", "unexplained tension at 18:03 with no preceding provocation visible"], ["lutek_day_burston_057", "tension appears at 18:03 and persists to 20:03 with no trigger shown"], ["lutek_day_burston_058", "tension builds through evening with no visible cause; accumulates oddly"], ["lutek_day_burston_059", "tension at 18:03 and 22:04 with no provoking event between"], ["lutek_day_burston_060", "seething at 20:03 after insult he let pass at 19:43; disproportionate"], ["lutek_day_burston_068", "lingers tense at 22:04 with no clear prior provocation visible"], ["lutek_day_burston_070", "residual temper at 22:04 after only a mild mock, no escalation shown"], ["lutek_day_burston_089", "tense at 18:00 answers curtly; temper lingers despite kindnesses"], ["lutek_multi_burstoff_014", "answers curtly to public mock; thick-skinned Lutek wouldn't snap"], ["lutek_multi_burstoff_026", "curt reply to mockery when tense; contradicts thick-skinned profile"], ["lutek_multi_burstoff_039", "curt reply to public mock contradicts thick-skinned profile"], ["lutek_multi_burston_003", "seething after mockery contradicts thick-skinned profile"], ["lutek_multi_burston_004", "seething after kindness; profile says mockery rolls off him"], ["lutek_multi_burston_014", "\"still seething, jaw tight\" after mock jars with thick-skinned profile"], ["lutek_multi_burston_017", "answers mock \"curtly and coldly\" contradicts thick-skinned profile"], ["lutek_multi_burston_025", "\"still seething\" after mild mockery; too intense for thick-skinned Lutek"], ["lutek_multi_burston_026", "\"curtly and coldly\" twice; sharper than profile allows for insults"], ["lutek_multi_burston_039", "answers mock curtly and coldly; profile says mockery rolls off"], ["welf_day_burston_098", "tense from 18:00 onward with no clear trigger, sleeps past midnight"], ["welf_multi_burston_019", "\"earlier temper\" lingers day 3 night with no visible trigger"], ["wojslaw_day_burstoff_008", "barbed mock at 07:58 draws no reaction; too mild for Wojslaw"], ["wojslaw_day_burstoff_016", "complains at kindnesses (15:24, 15:28); proud man grumbling at help jars"], ["wojslaw_day_burstoff_025", "lets public mockery pass at 08:52, then erupts later without new trigger"], ["wojslaw_day_burstoff_035", "erupts three times in evening including at 22:12 just before sleep"], ["wojslaw_day_burstoff_037", "lets three separate mocks pass with no reaction, too passive for profile"], ["wojslaw_day_burstoff_042", "lets public mockery pass at 20:55 after calm evening; too mild"], ["wojslaw_day_burstoff_046", "lets barbed remark pass at 10:11 and again at 17:48 while settled"], ["wojslaw_day_burstoff_048", "lets public mockery pass at 21:13 after a warm Marta-heavy evening"], ["wojslaw_day_burstoff_066", "lets first public mockery pass entirely, too lenient for Wojs\u0142aw"], ["wojslaw_day_burstoff_084", "ignores early mock then erupts later with no clear trigger escalation"], ["wojslaw_day_burstoff_098", "mocks him at 22:50 just before sleep; erupts then sleeps immediately, jarring close"], ["wojslaw_day_burston_008", "lets early public mockery pass with no reaction; jarring for Wojs\u0142aw"], ["wojslaw_day_burston_016", "mutters complaints at kindnesses (15:24, 15:28); prickly man but odd reaction"], ["wojslaw_day_burston_021", "erupts at 11:59, settled at 12:01 one minute later"], ["wojslaw_day_burston_026", "erupts at 21:47, settled at 22:04, erupts again at 22:32"], ["wojslaw_day_burston_035", "three separate fury eruptions same evening; saturation strains belief"], ["wojslaw_day_burston_037", "lets three mocks pass without reaction; too placid for Wojslaw"], ["wojslaw_day_burston_042", "proud man lets public mockery pass twice without reaction"], ["wojslaw_day_burston_046", "mocks ignored twice, prickly man too placid under insults"], ["wojslaw_day_burston_048", "erupts at midday mock, then ignores late-night mock when still tense"], ["wojslaw_day_burston_052", "erupts at late-night barb after calm well-fed evening; jarring spike"], ["wojslaw_day_burston_060", "erupts at 22:54 then sleeps settled at 23:02; too fast reset"], ["wojslaw_day_burston_099", "unusually warm all day; barely any irritability for this man"], ["wojslaw_multi_burstoff_097", "Day 3 12:07 erupts in fury at stranger offering kindness"], ["wojslaw_multi_burston_017", "Day 3 18:05 mutters complaint at a kindness, jarring"]], TMOD=[["test_acceptance_matrix", 23], ["test_authority", 8], ["test_behavioral_dynamics_validation", 17], ["test_believable_timescale", 5], ["test_burst", 31], ["test_burst_calibration", 18], ["test_burst_world", 4], ["test_calibration", 8], ["test_diagnostics", 1], ["test_expectations", 11], ["test_filters", 10], ["test_game_facing_longrun", 10], ["test_game_facing_memory", 15], ["test_game_facing_social_streams", 20], ["test_loss", 12], ["test_metrics", 7], ["test_narration_expression", 10], ["test_night", 6], ["test_order_invariance", 1], ["test_proactive_path", 11], ["test_resolution_factor", 5], ["test_social_event_mapper", 36], ["test_stability", 2], ["test_target_policy", 4], ["test_tick_golden", 27], ["test_time_scale", 3], ["test_weather_channel", 2]];
+  const FPER=[["wojslaw", 25], ["halgrim", 23], ["lutek", 23], ["branic", 19], ["cichy", 18], ["edda", 4], ["welf", 2]], FSUB=[["day burstON", 53], ["day burstOFF", 30], ["multi burstON", 24], ["multi burstOFF", 7]];
+  FPER.forEach(r=>document.getElementById('fper').insertAdjacentHTML('beforeend',`<tr><td>${r[0]}</td><td class="rate">${r[1]}</td></tr>`));
+  FSUB.forEach(r=>document.getElementById('fsub').insertAdjacentHTML('beforeend',`<tr><td>${r[0]}</td><td class="rate">${r[1]}</td></tr>`));
+  TMOD.forEach(r=>document.getElementById('tmod').insertAdjacentHTML('beforeend',`<tr><td class="mono">tests/${r[0]}.py</td><td class="rate">${r[1]}</td></tr>`));
+  function drawFlags(q){const t=document.getElementById('ftab');t.querySelectorAll('tr.f').forEach(e=>e.remove());
+    FLAGS.filter(f=>!q||(f[0]+f[1]).toLowerCase().includes(q)).forEach(f=>{const tr=document.createElement('tr');tr.className='f';tr.innerHTML=`<td class="mono" style="font-size:12px">${f[0]}</td><td class="sub" style="font-size:13px">${f[1]}</td>`;t.appendChild(tr);});}
+  drawFlags('');document.getElementById('fq').addEventListener('input',e=>drawFlags(e.target.value.toLowerCase().trim()));
+  </script>
+
   <div class="foot">
-    Full prose record: <code>eval/BELIEVABILITY_REPORT.md</code>. Raw final verdicts: <code>eval/phaseA_run5/</code>.
+    Full prose record: <code>eval/BELIEVABILITY_REPORT.md</code>. Raw final verdicts: <code>eval/phaseA_run5/</code>. Full per-scenario report (stimulus + narration per test): <code>eval/hourly_runs/TEST_REPORT.html</code>.
     Plans: <code>docs/plans/</code>. Engine state manual (inputs/effects + wiring graphs) lives in the private overlay repo.
   </div>