superlocalmemory 4.0.5 → 4.0.7

package/src/superlocalmemory/ui/js/od-brain.js

@@ -87,9 +87,16 @@
   }
 
   function phaseLabel(raw) {
-    return (raw || 'cold_start')
-      .replace(/_/g, '-')
-      .replace(/\b([a-z])/g, function (c) { return c.toUpperCase(); });
+    // Map known phase keys to human labels. Never string-mangle an internal
+    // identifier — unknown values fall back to a plain-English generic.
+    var LABELS = {
+      baseline:   'Baseline',
+      rule_based: 'Rule-based',
+      ml_model:   'ML model',
+      cold_start: 'Starting out',
+    };
+    var key = (raw || 'cold_start').toLowerCase().replace(/-/g, '_').replace(/\s+/g, '_');
+    return LABELS[key] || (raw ? String(raw).replace(/[-_]/g, ' ') : 'Starting out');
   }
 
   // ======================================================================
@@ -153,6 +160,35 @@
     return wrap;
   }
 
+  function humanizeSource(raw) {
+    // Map raw internal capability IDs to plain names a non-technical user can read.
+    // IDs follow patterns such as http:daemon-capability:<hex> or
+    // dashboard:local-capability:http-route:uid:<uid>:<hex>.
+    // Strip trailing hex hashes, then map the leading protocol/type prefix.
+    var s = String(raw || '')
+      .replace(/:?[0-9a-f]{40,}$/i, '')
+      .replace(/:?[0-9a-f]{32,}$/i, '')
+      .replace(/:$/, '')
+      .trim();
+    var prefix = (s.split(':')[0] || '').toLowerCase();
+    var MAP = {
+      http: 'Background service',
+      https: 'Background service',
+      dashboard: 'Dashboard',
+      cli: 'Command line',
+      'claude-code': 'Claude Code',
+      claude_code: 'Claude Code',
+      copilot: 'Copilot',
+      cursor: 'Cursor',
+      mcp: 'MCP server',
+      daemon: 'Background service',
+    };
+    if (MAP[prefix]) return MAP[prefix];
+    // Fall back: take the first segment, replace underscores/dashes, capitalise.
+    var label = prefix.replace(/[-_]/g, ' ').trim();
+    return label ? label.charAt(0).toUpperCase() + label.slice(1) : String(raw);
+  }
+
   function heatLegend() {
     var w = EL('div', { className: 'heat-legend' });
     w.appendChild(document.createTextNode('less '));
@@ -219,10 +255,10 @@
     var phaseNumber = Number(ranker.phase || 1);
     var modelActive = Boolean(ranker.model_active);
     var phaseDelta = modelActive
-      ? 'Verified active model'
+      ? 'Personalised model active'
       : signals < mlGate
-        ? fmtNum(mlGate - signals) + ' to ML data gate'
-        : 'ML data gate met · verified model required';
+        ? fmtNum(mlGate - signals) + ' more interactions to unlock personalisation'
+        : 'Ready to personalise — awaiting model verification';
     var healthStatus = (eng.health_status || 'INACTIVE').toUpperCase();
     var healthColor = healthStatus === 'HEALTHY' ? 'var(--ok)'
       : healthStatus === 'ACTIVE' ? 'var(--cyan)' : undefined;
@@ -250,17 +286,22 @@
     // Ranking phase: text label → isNumeric=false (font-size:24px to match design)
     strip.appendChild(kpiCard('skill', 'Ranking phase', phaseLabel(phase),
       phaseDelta, modelActive, undefined, false));
-    // Feedback signals: numeric → isNumeric=true
-    strip.appendChild(kpiCard('optimize', 'Feedback signals', fmtNum(signals),
-      '▲ ' + fmtNum(stats.unique_queries || 0) + ' unique queries', true, undefined, true));
+    // Questions answered: unique queries is the meaningful unit — total signals is context.
+    // 5,339 signals across 3 unique queries means one query repeated, not broad learning.
+    var uniqueQ = Number(stats.unique_queries || 0);
+    strip.appendChild(kpiCard('optimize', 'Questions answered', fmtNum(uniqueQ),
+      uniqueQ > 0
+        ? fmtNum(signals) + ' total interactions'
+        : 'no questions asked yet',
+      uniqueQ > 0, undefined, true));
     // Engagement health: text label → isNumeric=false
     strip.appendChild(kpiCard('health', 'Engagement health',
       healthStatus.charAt(0) + healthStatus.slice(1).toLowerCase(),
-      (eng.days_active || 0) + ' days active · ' + Number(eng.memories_per_day || 0).toFixed(1) + ' mem/day',
+      (eng.days_active || 0) + ' days active · ' + Number(eng.memories_per_day || 0).toFixed(1) + ' saves/day',
       healthStatus === 'HEALTHY', healthColor, false));
     // Patterns: numeric → isNumeric=true
     strip.appendChild(kpiCard('brain', 'Patterns learned', String(pCount),
-      '▲ ' + (beh.cross_project_transfers || 0) + ' transferable', pCount > 0, undefined, true));
+      (beh.cross_project_transfers || 0) + ' used across projects', pCount > 0, undefined, true));
     sec.appendChild(strip);
 
     // 2-column grid
@@ -276,8 +317,17 @@
     var pmeta = EL('div', {
       style: 'display:flex;justify-content:space-between;font-size:12px;color:var(--fg-2);margin-bottom:8px',
     });
-    pmeta.appendChild(EL('span', { text: fmtNum(signals) + ' / ' + fmtNum(mlGate) + ' signals' }));
-    pmeta.appendChild(EL('span', { className: 'num', text: pct + '%' }));
+    if (signals >= mlGate) {
+      // Gate already passed — showing the overrun fraction (e.g. "5,339 / 200") is
+      // misleading. Show the done state and what comes next instead.
+      pmeta.appendChild(EL('span', { text: fmtNum(signals) + ' interactions · gate passed' }));
+      var gdBadge = EL('span', { className: 'badge ' + (modelActive ? 'ok' : 'warn') });
+      gdBadge.appendChild(document.createTextNode(modelActive ? 'model active' : 'awaiting verification'));
+      pmeta.appendChild(gdBadge);
+    } else {
+      pmeta.appendChild(EL('span', { text: fmtNum(signals) + ' of ' + fmtNum(mlGate) + ' interactions' }));
+      pmeta.appendChild(EL('span', { className: 'num', text: pct + '%' }));
+    }
     pb.appendChild(pmeta);
     pb.appendChild(meter(pct));
     var phasesRow = EL('div', {
@@ -329,12 +379,12 @@
       ['Sources tracked',   String(stats.tracked_sources || 0)],
       ['Memory activity', truthCount(memoryActivity, 'facts_total', 'facts')],
       ['Feedback signals', truthCount(feedback, 'signals_total', 'signals')],
-      ['Claimed evidence', truthCount(experience, 'claimed_experiences_total', 'receipts')],
+      ['Claimed evidence', truthCount(experience, 'claimed_experiences_total', '')],
       ['Independently verified evidence', truthCount(
-        experience, 'independently_verified_experiences_total', 'receipts',
+        experience, 'independently_verified_experiences_total', '',
       )],
-      ['External observations', truthCount(externalEvidence, 'receipts_total', 'receipts')],
-      ['Correction quality', truthCount(correctionQuality, 'cases_total', 'review cases')],
+      ['External observations', truthCount(externalEvidence, 'receipts_total', '')],
+      ['Correction quality', truthCount(correctionQuality, 'cases_total', '')],
       ['Graph evidence', String(graph.fact_nodes || 0) + ' nodes · ' +
         String(graph.association_edges || 0) + ' edges'],
     ].forEach(function (row) {
@@ -367,26 +417,29 @@
       style: 'margin:0 0 14px;font-size:12px;line-height:1.55',
       text: 'SLM records completed work when an integration supplies evidence. These records do not change recall, ranking, or model routing by themselves.',
     }));
+    // Three plain-language KPI cards.
+    // Technical labels (Claimed evidence, Independently verified evidence,
+    // External observations, Correction quality) are preserved in the
+    // Privacy card rows above for test and audit traceability.
+    var claimedCount = Number(experience.claimed_experiences_total || 0);
+    var corrCount    = Number(correctionQuality.cases_total || 0);
+    var obsCount     = Number(externalEvidence.receipts_total || 0);
     var evGrid = EL('div', { className: 'kpi-strip', style: 'margin:0' });
-    evGrid.appendChild(kpiCard('fact_check', 'Claimed evidence',
-      truthCount(experience, 'claimed_experiences_total', ''), 'profile-scoped durable receipts',
-      Number(experience.claimed_experiences_total || 0) > 0, undefined, true));
-    evGrid.appendChild(kpiCard('verified', 'Independently verified evidence',
-      truthCount(experience, 'independently_verified_experiences_total', ''),
-      experience.verification_availability === 'not_supported_by_read_model'
-        ? 'not supported by this read model' : 'independent verifier result',
-      Number(experience.independently_verified_experiences_total || 0) > 0, undefined, true));
-    evGrid.appendChild(kpiCard('account_tree', 'Cognitive turns',
-      truthCount(experience, 'cognitive_turns_total', ''),
-      String((experience.cognitive_turns_by_state || {}).open || 0) + ' open · ' +
-        String((experience.cognitive_turns_by_state || {}).finalized || 0) + ' finalized',
-      Number(experience.cognitive_turns_total || 0) > 0, undefined, true));
-    evGrid.appendChild(kpiCard('account_tree', 'External observations',
+    evGrid.appendChild(kpiCard('fact_check', 'Tasks completed',
+      truthCount(experience, 'claimed_experiences_total', ''),
+      claimedCount > 0 ? 'work recorded from connected tools' : 'no tasks recorded yet',
+      claimedCount > 0, undefined, true));
+    evGrid.appendChild(kpiCard('tune', 'Corrections applied',
+      truthCount(correctionQuality, 'cases_total', ''),
+      corrCount > 0 ? 'improvements applied to results' : 'no corrections recorded yet',
+      corrCount > 0, undefined, true));
+    evGrid.appendChild(kpiCard('science', 'Observations logged',
       truthCount(externalEvidence, 'receipts_total', ''),
-      externalEvidence.availability === 'available'
-        ? String(externalEvidence.demonstrations_total || 0) + ' demonstrations · no automatic learning'
-        : 'external evidence unavailable',
-      Number(externalEvidence.receipts_total || 0) > 0, undefined, true));
+      externalEvidence.availability === 'available' &&
+        Number(externalEvidence.demonstrations_total || 0) > 0
+        ? String(externalEvidence.demonstrations_total) + ' demonstrations'
+        : 'no external demonstrations yet',
+      obsCount > 0, undefined, true));
     evb.appendChild(evGrid);
     evc.appendChild(evb);
     sec.appendChild(evc);
@@ -427,7 +480,7 @@
     hmhr.appendChild(EL('h3', { text: 'Reward signal density' }));
     hmhr.appendChild(EL('span', {
       className: 'sub',
-      text: 'settled numeric labels per day · last ' + Number(reward.window_days || 182) + ' days',
+      text: 'feedback events per day · last ' + Number(reward.window_days || 182) + ' days',
     }));
     hmhr.appendChild(EL('div', { className: 'spacer' }));
     hmhr.appendChild(heatLegend());
@@ -439,16 +492,26 @@
     hmcr.appendChild(hmbr);
     sec.appendChild(hmcr);
 
+    // Compute distribution state once so both cards use the same values.
+    var total = Number(reward.count || 0);
+    var bd = reward.distribution || {};
+    // isUnmeasuredPrior: when positive=0 and negative=0, every label sits on the
+    // initialization value (0.5). That is the absence of differentiation, not a
+    // finding — render it honestly rather than as a measured result.
+    var isUnmeasuredPrior = total > 0
+      && Number(bd.positive || 0) === 0
+      && Number(bd.negative || 0) === 0;
+
     // 2-column: sparkline + outcome mix
     var grid = EL('div', { className: 'grid', style: 'grid-template-columns:1fr 1fr;align-items:start' });
 
-    // Average settled reward and real daily series
+    // Recall quality card: average reward score + real daily sparkline
     var fbCard = EL('div', { className: 'card' });
     var fbH = EL('div', { className: 'card-head' });
-    fbH.appendChild(EL('h3', { text: 'Average settled reward' }));
+    fbH.appendChild(EL('h3', { text: 'Recall quality' }));
     fbH.appendChild(EL('span', {
       className: 'sub',
-      text: fmtNum(reward.count || 0) + ' finalized labels',
+      text: fmtNum(reward.count || 0) + (isUnmeasuredPrior ? ' interactions · default score' : ' interactions'),
     }));
     fbCard.appendChild(fbH);
     var fbB = EL('div', { className: 'card-pad' });
@@ -458,6 +521,13 @@
         style: 'font-size:30px;margin-bottom:12px',
         text: Number(reward.average).toFixed(3),
       }));
+      if (isUnmeasuredPrior) {
+        fbB.appendChild(EL('p', {
+          className: 'muted',
+          style: 'font-size:12px;margin-top:4px;margin-bottom:0',
+          text: 'Starting value — no differentiated engagement yet.',
+        }));
+      }
     }
     var fbSp = EL('div', { id: 'od-brain-sp-fb' });
     var sparkVals = timeline.slice(-30).map(function (point) {
@@ -471,7 +541,7 @@
       fbSp.appendChild(EL('p', {
         className: 'muted',
         style: 'padding:32px;text-align:center;font-size:13px',
-        text: 'No settled reward history is available yet.',
+        text: 'No recall history is available yet.',
       }));
     }
     fbB.appendChild(fbSp);
@@ -482,16 +552,17 @@
     var outCard = EL('div', { className: 'card' });
     var outH = EL('div', { className: 'card-head' });
     outH.appendChild(EL('h3', { text: 'Reward distribution' }));
-    outH.appendChild(EL('span', { className: 'sub', text: 'engagement-derived settled labels' }));
+    outH.appendChild(EL('span', { className: 'sub', text: 'based on how you engage with recalled results' }));
     outCard.appendChild(outH);
     var outB = EL('div', { className: 'card-pad', id: 'od-brain-outcomes' });
-    var total = Number(reward.count || 0);
-    var bd = reward.distribution || {};
-    if (total === 0) {
+    if (total === 0 || isUnmeasuredPrior) {
       outB.appendChild(EL('p', {
         className: 'muted',
         style: 'padding:16px;text-align:center;font-size:13px',
-        text: 'No settled reward labels yet. Recall engagement will populate this view.',
+        text: total === 0
+          ? 'No reward labels yet. Recall engagement will populate this view.'
+          : 'All ' + fmtNum(total) + ' labels carry the default score. ' +
+            'Differentiated results appear after consistent recall use.',
       }));
     } else {
       [
@@ -530,7 +601,7 @@
     var tc = EL('div', { className: 'card' });
     var tch = EL('div', { className: 'card-head' });
     tch.appendChild(EL('h3', { text: 'Tech preferences' }));
-    tch.appendChild(EL('span', { className: 'sub', text: 'Layer 1 · confidence-weighted' }));
+    tch.appendChild(EL('span', { className: 'sub', text: 'your tools and technology preferences' }));
     tc.appendChild(tch);
     var tcb = EL('div', { className: 'card-pad' });
     var techItems = l.tech_preferences || [];
@@ -571,7 +642,7 @@
     var wc = EL('div', { className: 'card' });
     var wch = EL('div', { className: 'card-head' });
     wch.appendChild(EL('h3', { text: 'Workflow patterns' }));
-    wch.appendChild(EL('span', { className: 'sub', text: 'Layer 3 · sequence & temporal' }));
+    wch.appendChild(EL('span', { className: 'sub', text: 'how you sequence your work over time' }));
     wc.appendChild(wch);
     var wcb = EL('div', { className: 'card-pad' });
     var wfPats = l.workflow_patterns || [];
@@ -672,11 +743,26 @@
   // ======================================================================
   // Tab: CONNECTED CLIENTS
   // ======================================================================
-  function buildClients(living, configured) {
+  function fmtSecondsAgo(s) {
+    if (s == null) return 'unknown';
+    var n = Number(s);
+    if (n < 120) return n + 's ago';
+    if (n < 7200) return Math.round(n / 60) + 'm ago';
+    if (n < 172800) return Math.round(n / 3600) + 'h ago';
+    return Math.round(n / 86400) + 'd ago';
+  }
+
+  // buildClients accepts an optional third argument `boundedLoops` (brain.bounded_loops).
+  // Absent / undefined is safe: the BL section is silently omitted.
+  function buildClients(living, configured, boundedLoops) {
     var sec = EL('section', { className: 'tabpane', 'data-p': 'clients' });
-    var clients = ((living && living.connected_clients) || {}).clients || [];
+    var connectedData = (living && living.connected_clients) || {};
+    var clients = connectedData.clients || [];
+    var registryStatus = String(connectedData.registry_status || '');
+    var newestAgo = connectedData.newest_entry_seconds_ago;
     var configuredData = configured || {};
 
+    // ── Recent client activity ────────────────────────────────────────────
     // Activity is presence reported by host lifecycle hooks, not the old
     // tool-event proxy.  A configured adapter and a recent client are two
     // different truths, rendered as separate cards.
@@ -686,8 +772,46 @@
     evh.appendChild(EL('span', { className: 'sub', text: 'host lifecycle presence · last 5 minutes' }));
     evc.appendChild(evh);
     var evb = EL('div', { className: 'card-pad' });
-    if (clients.length === 0) {
-      evb.appendChild(EL('p', { className: 'muted', style: 'padding:16px;text-align:center',
+
+    if (registryStatus === 'error' || registryStatus === 'unknown') {
+      // Wave 4 honesty rule: failure and emptiness must not return the same value.
+      // 'unknown' belongs HERE, not in the healthy-empty branch below. The
+      // registry reader is fail-soft — a corrupt or unreadable file makes
+      // _load() return {} and active_client_summary() return [], so the read
+      // model reports status 'unknown' with an empty client list and is_real
+      // true. Falling through to "No host activity in the last 5 minutes" would
+      // render a BROKEN registry exactly like a healthy quiet machine, which is
+      // precisely the silent failure this section was rewritten to eliminate.
+      evb.appendChild(EL('p', {
+        className: 'muted',
+        style: 'padding:16px;text-align:center;font-size:13px',
+        text: registryStatus === 'unknown'
+          ? 'Presence records could not be read, so recent activity cannot be '
+            + 'confirmed. This is not the same as no agents being active.'
+          : 'Presence registry unavailable. Check daemon logs for details.',
+      }));
+    } else if (registryStatus === 'stale') {
+      // TELEMETRY GAP — entries exist but all are > 10 min old.
+      // "No activity" here would be indistinguishable from "hooks not firing".
+      evb.appendChild(EL('p', {
+        className: 'muted',
+        style: 'padding:16px;text-align:center;font-size:13px',
+        text: 'Presence recording gap — last hook event was ' + fmtSecondsAgo(newestAgo) + '. ' +
+          'Hooks are installed but presence has not been written recently. ' +
+          'Recent activity from integrations cannot be confirmed.',
+      }));
+    } else if (registryStatus === 'absent' || registryStatus === 'empty') {
+      // No records at all — first run or hooks have not fired yet.
+      evb.appendChild(EL('p', {
+        className: 'muted',
+        style: 'padding:16px;text-align:center;font-size:13px',
+        text: 'No presence records found. Hook events may not have fired yet on this install.',
+      }));
+    } else if (clients.length === 0) {
+      // Registry is live (recent writes) but no clients in the 5-min window.
+      evb.appendChild(EL('p', {
+        className: 'muted',
+        style: 'padding:16px;text-align:center;font-size:13px',
         text: 'No host activity in the last 5 minutes. This does not mean an integration is uninstalled.',
       }));
     } else {
@@ -695,7 +819,7 @@
         var row = EL('div', { className: 'list-row' });
         row.appendChild(EL('b', { style: 'flex:1', text: String(client.kind || 'other') }));
         row.appendChild(EL('span', { className: 'muted',
-          text: 'active ' + Number(client.last_seen_seconds_ago || 0) + 's ago',
+          text: 'active ' + fmtSecondsAgo(client.last_seen_seconds_ago),
         }));
         evb.appendChild(row);
       });
@@ -703,8 +827,11 @@
     evc.appendChild(evb);
     sec.appendChild(evc);
 
-    // Configured integrations are installation/sync state, not client activity.
-    var tc = EL('div', { className: 'card' });
+    // ── Configured integrations ───────────────────────────────────────────
+    // Installation / sync state, not client activity.
+    // For Codex, evidence_tier='configured' means config-file evidence;
+    // badge reads 'configured' rather than 'available' to be explicit.
+    var tc = EL('div', { className: 'card', style: 'margin-bottom:16px' });
     var tch = EL('div', { className: 'card-head' });
     tch.appendChild(EL('h3', { text: 'Configured integrations' }));
     tch.appendChild(EL('span', { className: 'sub', text: 'installation and sync availability' }));
@@ -714,14 +841,50 @@
       var state = configuredData[kind] || {};
       var row = EL('div', { className: 'list-row' });
       row.appendChild(EL('span', { style: 'flex:1', text: kind.replace(/_/g, ' ') }));
-      row.appendChild(EL('span', {
+      var badgeText = state.active
+        ? (state.evidence_tier === 'configured' ? 'configured' : 'available')
+        : (state.reason || 'not available');
+      var badgeEl = EL('span', {
         className: 'badge ' + (state.active ? 'ok' : 'warn'),
-        text: state.active ? 'available' : (state.reason || 'not available'),
-      }));
+        text: badgeText,
+      });
+      if (state.evidence) badgeEl.setAttribute('title', String(state.evidence));
+      row.appendChild(badgeEl);
       tcb.appendChild(row);
     });
     tc.appendChild(tcb);
     sec.appendChild(tc);
+
+    // ── Bounded Loops ─────────────────────────────────────────────────────
+    // Section is on when bounded_loops.section_enabled === true.
+    // When absent (older API or not installed) the section is silently omitted.
+    var bl = boundedLoops || {};
+    if (bl.section_enabled) {
+      var blc = EL('div', { className: 'card' });
+      var blh = EL('div', { className: 'card-head' });
+      blh.appendChild(EL('h3', { text: 'Bounded Loops' }));
+      blh.appendChild(EL('span', {
+        className: 'sub',
+        text: 'local installation' + (bl.version ? ' · v' + bl.version : ''),
+      }));
+      blc.appendChild(blh);
+      var blb = EL('div', { className: 'card-pad', style: 'display:flex;flex-direction:column;gap:2px' });
+      [
+        ['Status', 'installed'],
+        ['Version', bl.version || 'unknown'],
+        ['Bridge contract', bl.bridge_contract || '—'],
+        ['Evidence', bl.evidence || bl.evidence_tier || '—'],
+        ['Note', bl.note || '—'],
+      ].forEach(function (r) {
+        var row = EL('div', { className: 'list-row' });
+        row.appendChild(EL('span', { className: 'muted', style: 'flex:1', text: r[0] }));
+        row.appendChild(EL('b', { text: String(r[1]) }));
+        blb.appendChild(row);
+      });
+      blc.appendChild(blb);
+      sec.appendChild(blc);
+    }
+
     return sec;
   }
 
@@ -737,7 +900,7 @@
     var card = EL('div', { className: 'card' });
     var ch = EL('div', { className: 'card-head' });
     ch.appendChild(EL('h3', { text: 'Source quality' }));
-    ch.appendChild(EL('span', { className: 'sub', text: 'persisted source-outcome posterior · 0.0–1.0' }));
+    ch.appendChild(EL('span', { className: 'sub', text: 'how well each source has performed · 0.0–1.0' }));
     card.appendChild(ch);
     var cb = EL('div', { className: 'card-pad' });
     var aggregate = (living && living.source_quality) || {};
@@ -752,27 +915,44 @@
       cb.appendChild(EL('p', {
         className: 'muted',
         style: 'padding:16px;text-align:center;font-size:13px',
-        text: String(observedSources) + ' sources have evidence, but no per-source posterior is available yet.',
+        text: String(observedSources) + ' sources have been seen, but individual quality scores are not available yet.',
       }));
     } else {
-      cb.appendChild(EL('p', {
-        className: 'muted',
-        style: 'font-size:13px;margin-bottom:16px',
-        text: String(observedSources) + ' sources with observed mean quality ' +
-          Number(aggregate.mean_quality).toFixed(3) + '.',
-      }));
-      entries.forEach(function (k) {
-        var v = Number(scores[k]);
-        var row = EL('div', { style: 'margin-bottom:14px' });
-        var meta = EL('div', {
-          style: 'display:flex;justify-content:space-between;font-size:13px;margin-bottom:6px',
-        });
-        meta.appendChild(EL('span', { className: 'mono', text: k }));
-        meta.appendChild(EL('b', { className: 'num', text: v.toFixed(2) }));
-        row.appendChild(meta);
-        row.appendChild(meter(v * 100));
-        cb.appendChild(row);
+      // allAtPrior: every source at exactly 0.50 means the system has not yet accumulated
+      // enough outcome data to shift away from the initialization value. Showing 18 identical
+      // rows would mislead a non-technical reader into thinking quality was measured.
+      var allAtPrior = entries.every(function (k) {
+        return Math.abs(Number(scores[k]) - 0.5) < 0.005;
       });
+      if (allAtPrior) {
+        cb.appendChild(EL('p', {
+          className: 'muted',
+          style: 'font-size:13px;margin-bottom:16px',
+          text: String(observedSources) + ' source' + (observedSources === 1 ? '' : 's') +
+            ' observed. Quality scores start at 0.5 and shift once recall shows ' +
+            'which sources consistently produce better results. No quality signal has settled yet.',
+        }));
+      } else {
+        cb.appendChild(EL('p', {
+          className: 'muted',
+          style: 'font-size:13px;margin-bottom:16px',
+          text: String(observedSources) + ' source' + (observedSources === 1 ? '' : 's') +
+            ' · average quality ' + Number(aggregate.mean_quality).toFixed(2) + '.',
+        }));
+        entries.forEach(function (k) {
+          var v = Number(scores[k]);
+          var displayName = humanizeSource(k);
+          var row = EL('div', { style: 'margin-bottom:14px' });
+          var meta = EL('div', {
+            style: 'display:flex;justify-content:space-between;font-size:13px;margin-bottom:6px',
+          });
+          meta.appendChild(EL('span', { text: displayName }));
+          meta.appendChild(EL('b', { className: 'num', text: v.toFixed(2) }));
+          row.appendChild(meta);
+          row.appendChild(meter(v * 100));
+          cb.appendChild(row);
+        });
+      }
     }
     card.appendChild(cb); sec.appendChild(card);
     return sec;
@@ -880,7 +1060,7 @@
         buildOverview(learning, behavioral, dateMap, living),
         buildReward(behavioral),
         buildBehaviour(learning, behavioral),
-        buildClients(living, brain.cross_platform),
+        buildClients(living, brain.cross_platform, brain.bounded_loops),
         buildSourceQuality(learning, living)
       );
       wireTabs(container);