@cachly-dev/mcp-server 0.10.65 → 0.10.66

package/src/handlers/brain.ts

@@ -1,2622 +0,0 @@
-import { execSync } from 'node:child_process';
-import type { Redis } from 'ioredis';
-import { calculateConfidence, confidenceBadge, STRUCTURED_TEMPLATES,
-         CONFIDENCE_WARN_VALUE, CONFIDENCE_STALE_VALUE, CONFIDENCE_WARN_DAYS } from '../confidence.js';
-import { ckgSlug, extractProblemConcept, ckgUpsertNode, ckgUpdateEdge,
-         ckgUpsertPersonNode, ckgUpsertFileNode } from '../ckg.js';
-import { safeJsonParse, scanKeys } from '../utils.js';
-import type { CKGEdge, CKGNode, PersonNode } from '../ckg.js';
-import { keywordSearch, tokenize, splitMultiQuery, levenshtein,
-         indexVocab as _indexVocab } from '../search.js';
-import { rerankByQuality } from '../rerank.js';
-import { computeEmbedding, hasEmbedProvider } from '../embeddings.js';
-
-// ── Changelog (shown once per version in session_start) ──────────────────────
-const MCP_VERSION = '0.10.65';
-const WHATS_NEW: Record<string, string[]> = {
-  '0.10.64': [
-    `📈 **The three decisive metrics, now measurable**`,
-    `  ⏱️ \`brain_metrics()\` — time-to-first-recall, recall-lift, team-knowledge-reuse in one view`,
-    `  👥 Cross-author reuse tracked: recall a teammate's lesson → counted + surfaced inline`,
-    `  🛡️ Stability: scans now capped + timed out (a huge keyspace can't hang the agent)`,
-  ],
-  '0.10.61': [
-    `🎯 **Phase 3C: 100 MCP tools milestone — zero-setup knowledge graph**`,
-    `  🔄 \`brain_from_git\` now auto-builds Person+File nodes from git history (zero setup!)`,
-    `  🔍 \`skill_gaps()\` — find domains with unresolved failures, missing attribution`,
-    `  📊 \`brain_coverage()\` — scored 0-100 health report: lessons · attribution · file coverage`,
-  ],
-  '0.10.60': [
-    `🆕 **Phase 3B: File map + team overview + visibility**`,
-    `  📁 \`brain_file_map(file_paths=[...])\` — experts + lessons per file before you touch it`,
-    `  🗺️ \`team_expertise_map()\` — full team skills matrix in one table`,
-    `  🔒 \`visibility: "private"\` on \`learn_from_attempts\` — private notes never leak to smart_recall`,
-    `  📊 98 MCP tools`,
-  ],
-  '0.10.59': [
-    `🆕 **Phase 3A: Org-wide knowledge graph**`,
-    `  👥 \`brain_who_knows(topic="...")\` — find your team's experts on any topic instantly`,
-    `  🕸️ Person + File nodes auto-built from \`learn_from_attempts(author="...", file_paths=[...])\``,
-    `  👤 Author attribution now shown inline in \`smart_recall\` results`,
-    `  📊 96 MCP tools · 386 tests green`,
-  ],
-  '0.10.58': [
-    `🧹 Zero lint warnings — all unused imports cleaned across every handler`,
-    `✅ Build: 0 errors, 0 warnings · 379 tests green`,
-  ],
-  '0.10.57': [
-    `🆕 **What's new in v${MCP_VERSION}:**`,
-    `  🔧 **Critical fix** — corrected the npm bin/main entry path (\`dist/src/index.js\`); 0.10.50–0.10.52 shipped a broken entry point`,
-    `  🔧 Version hygiene — synced package.json, server.json, lockfile, MCP_VERSION; \`latest\` now points forward again`,
-    `  ✅ Accurate tool count (95) across README, server.json, CLI banners`,
-    `  ✅ Includes all Phase 2/3 work: quality reranking, Cachly-Bench, hybrid + CKG recall, team_confirm governance`,
-  ],
-  '0.10.52': [
-    `  ✅ \`smart_recall\` — CKG traversal as Layer 3: finds lessons that FIXED causal-graph-similar problems (🕸️ causal graph badge)`,
-    `  ✅ Bench corpus expanded: governance adversarial pair proves review-boost end-to-end (+22% P@1 headline)`,
-    `  🔧 Import cleanup: unused symbols removed from brain.ts (no functional change)`,
-    `  💡 \`smart_recall\` now shows keyword / semantic / CKG / hybrid match type for every result`,
-  ],
-  '0.10.51': [
-    `  ✅ \`team_confirm\` — human review raises lesson recall ranking (🛡️ senior / ✔️ peer)`,
-    `  ✅ Governance-aware recall — confirmed lessons outrank unreviewed auto-learned entries`,
-  ],
-  '0.10.50': [
-    `  ✅ \`smart_recall\` — unified keyword + semantic hybrid list (one ranked result, not two sections)`,
-    `  ✅ \`smart_recall\` — "Brain saved you here" banner surfaces time saved inline per proven lesson`,
-    `  ✅ Quality reranking — proven successes outrank symptom-dense failures (+11% Precision@1)`,
-    `  ✅ Cachly-Bench — IR benchmark with CI regression guard (see BENCH.md for the proof)`,
-    `  🔧 Contradiction resolution now persisted to Redis (90-day TTL)`,
-    `  💡 Run \`smart_recall\` to see your Brain's recall quality in action`,
-  ],
-  '0.10.49': [
-    `  ✅ \`brain_from_git\` — auto-seed your Brain from git history in seconds`,
-    `  ✅ \`brain_predict_failures\` — predict CI/build failures before they happen`,
-    `  ✅ VS Code extension — Brain status bar, ambient learning, CodeLens hints`,
-    `  ✅ Team Brain — share lessons across your whole engineering team`,
-    `  🔧 Stability — auto-expiring memory keys, longer provisioning wait, robust recall`,
-  ],
-};
-
-// Shared types used in brain handlers
-export interface Instance {
-  id: string; name: string; tier: string; status: string; region: string;
-  host?: string; port?: number; password?: string; tls_enabled?: boolean;
-  vector_token?: string; memory_mb: number; encryption_at_rest: boolean;
-  created_at: string;
-}
-
-interface SemanticSearchResponse {
-  found: boolean;
-  id?: string;
-  similarity?: number;
-  prompt?: string;
-}
-
-type GetConnection = (instanceId: string) => Promise<Redis>;
-type ApiFetch = <T>(path: string, options?: RequestInit) => Promise<T>;
-
-export const BRAIN_TOOL_NAMES = new Set([
-  'learn_from_attempts', 'recall_best_solution', 'smart_recall',
-  'session_start', 'session_end', 'session_ping', 'session_handoff', 'auto_learn_session',
-  'brain_who_knows', 'brain_file_map', 'team_expertise_map',
-  'skill_gaps', 'brain_coverage', 'brain_metrics',
-]);
-
-export async function handleBrainTool(
-  name: string,
-  args: Record<string, unknown>,
-  getConnection: GetConnection,
-  apiFetch: ApiFetch,
-): Promise<string | null> {
-  switch (name) {
-    case 'learn_from_attempts': {
-      const {
-        instance_id,
-        topic,
-        outcome,
-        what_worked,
-        what_failed = '',
-        context: ctx = '',
-        severity = 'major',
-        file_paths = [],
-        commands = [],
-        tags = [],
-        depends_on = [],
-        author = '',
-        visibility = 'team',
-      } = args as {
-        instance_id: string;
-        topic: string;
-        outcome: 'success' | 'failure' | 'partial';
-        what_worked: string;
-        what_failed?: string;
-        context?: string;
-        severity?: 'critical' | 'major' | 'minor';
-        file_paths?: string[];
-        commands?: string[];
-        tags?: string[];
-        depends_on?: string[];
-        author?: string;
-        visibility?: 'public' | 'team' | 'private';
-      };
-
-      const redis = await getConnection(instance_id);
-      const ts = new Date().toISOString();
-
-      // ── Structured template hints ──────────────────────────────────────────
-      const category = topic.split(':')[0];
-      const template = STRUCTURED_TEMPLATES[category];
-      const templateWarnings: string[] = [];
-      if (template) {
-        for (const req of template.required) {
-          if (req === 'commands' && commands.length === 0) {
-            templateWarnings.push(`📋 ${template.hint}`);
-          }
-        }
-      }
-
-      // ── Deduplication + audit trail ────────────────────────────────────────
-      let isUpdate = false;
-      let recallCount = 0;
-      let auditTrail: Array<{ ts: string; action: string; prev_outcome?: string }> = [];
-      const existingRaw = await redis.get(`cachly:lesson:best:${topic}`);
-      if (existingRaw) {
-        try {
-          const prev = JSON.parse(existingRaw) as {
-            recall_count?: number;
-            outcome?: string;
-            audit_trail?: Array<{ ts: string; action: string; prev_outcome?: string }>;
-          };
-          recallCount = prev.recall_count ?? 0;
-          auditTrail = prev.audit_trail ?? [];
-          auditTrail.push({ ts, action: 'updated', prev_outcome: prev.outcome });
-          if (auditTrail.length > 20) auditTrail = auditTrail.slice(-20);
-          isUpdate = true;
-
-          // ── Contradiction detection ─────────────────────────────────────────
-          const contradictionWarning: string[] = [];
-          if (prev.outcome === 'success' && outcome === 'failure') {
-            contradictionWarning.push(
-              `⚠️ **Contradiction detected!** Existing lesson has outcome: \`success\`, but you're storing \`failure\`.`,
-              `The existing "success" lesson will be preserved. Only the audit trail is updated.`,
-              `If you meant to mark this as failed permanently, store a new lesson with a distinct topic slug.`,
-            );
-          } else if (prev.outcome === 'failure' && outcome === 'success') {
-            contradictionWarning.push(
-              `✅ **Conflict resolved!** Previous lesson was \`failure\` — now overwriting with \`success\`.`,
-            );
-          }
-          if (contradictionWarning.length > 0) {
-            // Store contradiction audit but don't block
-            auditTrail[auditTrail.length - 1].action = 'contradiction-resolved';
-            // Layer 3: Write CKG contradicts edge for MADC to process
-            try {
-              const cId = ckgSlug(topic);
-              const resId = ckgSlug(`resolution:${topic}`);
-              await ckgUpdateEdge(redis, cId, 'contradicts', resId, false);
-            } catch { /* non-critical */ }
-            // Persist a queryable resolution record — so contradiction history survives
-            // (CKG edges alone aren't easily auditable). Bounded + TTL'd like other history.
-            try {
-              const resolutionKey = `cachly:contradictions:${topic}`;
-              const direction = prev.outcome === 'success' && outcome === 'failure'
-                ? 'kept-success' : 'overwrote-failure';
-              await redis.rpush(resolutionKey, JSON.stringify({
-                ts, prev_outcome: prev.outcome, new_outcome: outcome, direction,
-              }));
-              await redis.ltrim(resolutionKey, -50, -1);
-              await redis.expire(resolutionKey, 180 * 86400);
-            } catch { /* non-critical */ }
-            contradictionWarning.push(`🗳️ Run \`madc_deliberate(topic="${topic}")\` to resolve via expert agent voting.`);
-          }
-        } catch { /* ignore parse error */ }
-      } else {
-        auditTrail = [{ ts, action: 'created' }];
-      }
-
-      // ── "I Was Wrong" Protocol — failure attribution ───────────────────────
-      const iWasWrongWarning: string[] = [];
-      if (outcome === 'failure') {
-        // Search for related success lessons that might have prevented this failure
-        const scanKeys: string[] = [];
-        const scanStream = redis.scanStream({ match: 'cachly:lesson:best:*', count: 100 });
-        await new Promise<void>((res, rej) => {
-          scanStream.on('data', (b: string[]) => scanKeys.push(...b));
-          scanStream.on('end', res);
-          scanStream.on('error', rej);
-        });
-        const topicWords = topic.split(/[:\-_]/).filter(w => w.length > 2);
-        for (const k of scanKeys.slice(0, 50)) {
-          const raw = await redis.get(k);
-          if (!raw) continue;
-          try {
-            const l = JSON.parse(raw) as { outcome?: string; topic?: string; severity?: string };
-            if (l.outcome !== 'success') continue;
-            const lWords = (l.topic ?? '').split(/[:\-_]/).filter(w => w.length > 2);
-            const overlap = topicWords.filter(w => lWords.includes(w)).length;
-            if (overlap >= 1 && l.topic !== topic) {
-              iWasWrongWarning.push(
-                `⚠️ **"I Was Wrong"**: lesson \`${l.topic}\` (success, ${l.severity ?? 'major'}) might have prevented this failure.`,
-                `   → Use \`recall_best_solution(topic="${l.topic}")\` before next attempt.`,
-                `   → To mark it critical: \`learn_from_attempts(topic="${l.topic}", ..., severity="critical")\``,
-              );
-              break; // only show the most relevant match
-            }
-          } catch { /* skip */ }
-        }
-      }
-
-      // ── Register dependency index for causal chain ─────────────────────────
-      for (const dep of depends_on) {
-        const depKey = `cachly:dep:${dep}`;
-        const existing = await redis.get(depKey);
-        const depTopics: string[] = safeJsonParse<string[]>(existing, []);
-        if (!depTopics.includes(topic)) depTopics.push(topic);
-        await redis.set(depKey, JSON.stringify(depTopics), 'EX', 90 * 86400);
-      }
-
-      const lessonObj = {
-        topic,
-        outcome,
-        what_worked,
-        what_failed,
-        context: ctx,
-        severity,
-        file_paths,
-        commands,
-        tags,
-        depends_on,
-        ...(author ? { author } : {}),
-        visibility,
-        recall_count: recallCount,
-        ts,
-        verified_at: outcome === 'success' || outcome === 'partial' ? ts : undefined,
-        confidence: 1.0,
-        audit_trail: auditTrail,
-        version: 3,
-      };
-      const lesson = JSON.stringify(lessonObj);
-
-      // Metric 1 (time-to-first-recall): mark the moment the Brain first gained
-      // knowledge. SET NX so only the very first learn wins. Fire-and-forget.
-      redis.set(`cachly:stats:born_at:${instance_id}`, ts, 'EX', 365 * 86400, 'NX').catch(() => {});
-
-      // Always append to the history list (audit log); keep last 100 entries, 90-day TTL
-      const listKey = `cachly:lessons:${topic}`;
-      await redis.rpush(listKey, lesson);
-      await redis.ltrim(listKey, -100, -1);
-      await redis.expire(listKey, 90 * 86400);
-
-      // Update best key for success/partial; for failure only update if no success exists
-      if (outcome === 'success' || outcome === 'partial') {
-        await redis.set(`cachly:lesson:best:${topic}`, lesson);
-      } else if (!existingRaw) {
-        await redis.set(`cachly:lesson:best:${topic}`, lesson);
-      }
-
-      // Track in decision log for session replay
-      try {
-        const dlKey = 'cachly:session:decision-log';
-        const dlEntry = JSON.stringify({ ts, topic, outcome, what_worked: what_worked.slice(0, 120) });
-        await redis.rpush(dlKey, dlEntry);
-        await redis.ltrim(dlKey, -50, -1);
-      } catch { /* non-critical */ }
-
-      // ── Layer 1+2: CKG update (Causal Knowledge Graph + Belief Update Engine) ──
-      // BUE: Bayesian confidence, contradiction detection, second-degree propagation, decay
-      let beliefConflict: string | null = null;
-      try {
-        const conceptId = ckgSlug(topic);
-        const domain = topic.split(':')[0] ?? 'unknown';
-        const conceptType = domain; // fix, debug, deploy, infra, api, etc.
-
-        // Upsert concept node
-        await ckgUpsertNode(redis, conceptId, domain, conceptType);
-
-        // Tag co-occurrence edges
-        for (const tag of tags) {
-          const tagId = ckgSlug(`tag:${tag}`);
-          await ckgUpsertNode(redis, tagId, 'tag', 'tag');
-          await ckgUpdateEdge(redis, conceptId, 'co-occurs', tagId, outcome === 'success', outcome === 'partial');
-        }
-
-        // depends_on → requires edges (structural, always confidence 1.0 direction)
-        for (const dep of depends_on) {
-          const depId = ckgSlug(dep);
-          await ckgUpdateEdge(redis, conceptId, 'requires', depId, true);
-        }
-
-        // fixes edge: if category=fix and outcome=success, link to problem concept
-        if ((domain === 'fix' || domain === 'debug') && (outcome === 'success' || outcome === 'partial')) {
-          const problemText = what_failed || ctx || '';
-          const problemConcept = problemText ? extractProblemConcept(problemText) : null;
-          if (problemConcept) {
-            const problemId = ckgSlug(`problem:${problemConcept}`);
-            await ckgUpsertNode(redis, problemId, 'problem', 'problem');
-            await ckgUpdateEdge(redis, conceptId, 'fixes', problemId, outcome === 'success', outcome === 'partial');
-          }
-        }
-
-        // causes edge: if outcome=failure, link topic concept to the problem context
-        if (outcome === 'failure' && (what_failed || what_worked)) {
-          const causeText = what_failed || what_worked;
-          const causeConcept = extractProblemConcept(causeText);
-          if (causeConcept) {
-            const causeId = ckgSlug(`cause:${causeConcept}`);
-            await ckgUpsertNode(redis, causeId, 'cause', 'cause');
-            await ckgUpdateEdge(redis, conceptId, 'causes', causeId, false);
-          }
-        }
-
-        // ── BUE: Contradiction detection ──────────────────────────────────────
-        // If this topic previously had a confirmed 'fixes' edge (confidence > 0.7)
-        // and now outcome=failure → flag belief_conflict
-        if (outcome === 'failure') {
-          const existingEdgeKeys = await redis.smembers(`cachly:ckg:idx:from:${conceptId}`);
-          for (const ek of existingEdgeKeys) {
-            const er = await redis.get(ek);
-            if (!er) continue;
-            const existEdge = safeJsonParse<CKGEdge | null>(er, null);
-            if (!existEdge) continue;
-            if (existEdge.edgeType === 'fixes' && existEdge.confidence > 0.7 && existEdge.trials >= 3) {
-              beliefConflict = `⚠️ **belief_conflict** on \`${topic}\`: previously confirmed fix (confidence ${existEdge.confidence.toFixed(2)}, n=${existEdge.trials}) now reports failure. Both beliefs retained as \`contested\`. Use \`ckg_inspect(concept="${conceptId}")\` to review.`;
-              // Store conflict marker
-              const conflictKey = `cachly:ckg:conflict:${conceptId}`;
-              await redis.set(conflictKey, JSON.stringify({ topic, detected_at: new Date().toISOString(), fix_confidence: existEdge.confidence, fix_trials: existEdge.trials, failure_outcome: outcome }), 'EX', 60 * 60 * 24 * 90);
-              // Auto-trigger MADC deliberation in background — never blocks learn_from_attempts
-              // Note: fire-and-forget; madc_deliberate is handled elsewhere in the switch
-              // (conflict marker stored above — no warm-up call needed)
-            }
-          }
-        }
-
-        // ── BUE: Second-degree propagation ────────────────────────────────────
-        // When a 'fixes' edge gets stronger, boost co-occurring second-degree edges slightly
-        if (outcome === 'success' && (domain === 'fix' || domain === 'debug')) {
-          const fromEdgeKeys = await redis.smembers(`cachly:ckg:idx:from:${conceptId}`);
-          for (const ek of fromEdgeKeys.slice(0, 10)) {
-            const er = await redis.get(ek);
-            if (!er) continue;
-            const e2 = safeJsonParse<CKGEdge | null>(er, null);
-            if (!e2) continue;
-            if (e2.edgeType !== 'fixes') continue;
-            // Boost second-degree: edges from e2.to get a small fractional success
-            const secondKeys = await redis.smembers(`cachly:ckg:idx:from:${e2.to}`);
-            for (const sk of secondKeys.slice(0, 5)) {
-              const sr = await redis.get(sk);
-              if (!sr) continue;
-              const se = safeJsonParse<CKGEdge | null>(sr, null);
-              if (!se) continue;
-              if (se.edgeType !== 'co-occurs') continue;
-              // Add 0.1 fractional success (second-degree signal)
-              se.successes = (se.successes || 0) + 0.1;
-              se.trials = (se.trials || 0) + 0.1;
-              se.confidence = (se.successes + 1) / (se.trials + 2);
-              se.last_updated = new Date().toISOString();
-              await redis.set(sk, JSON.stringify(se));
-            }
-          }
-        }
-
-        // ── BUE: Stale edge decay ─────────────────────────────────────────────
-        // Edges older than 90 days with < 3 trials decay by 10% confidence.
-        // Only run probabilistically (1% of calls) to avoid per-call overhead.
-        if (Math.random() < 0.01) {
-          const decayCutoff = Date.now() - 90 * 24 * 60 * 60 * 1000;
-          const fromEdgeKeys = await redis.smembers(`cachly:ckg:idx:from:${conceptId}`);
-          for (const ek of fromEdgeKeys) {
-            const er = await redis.get(ek);
-            if (!er) continue;
-            const de = safeJsonParse<CKGEdge | null>(er, null);
-            if (!de) continue;
-            if (de.trials < 3 && de.last_updated && new Date(de.last_updated).getTime() < decayCutoff) {
-              de.confidence = de.confidence * 0.9;
-              de.last_updated = new Date().toISOString();
-              await redis.set(ek, JSON.stringify(de));
-            }
-          }
-        }
-      } catch { /* CKG updates are non-critical */ }
-
-      // ── Phase 3A: People + File nodes in knowledge graph ──────────────────────
-      // Builds the "who knows what" map automatically from author + file_paths.
-      // Person → authored → Concept edges power brain_who_knows queries.
-      if (author) {
-        try {
-          const domain = topic.split(':')[0] ?? 'unknown';
-          const personId = await ckgUpsertPersonNode(redis, author, domain);
-          const conceptId = ckgSlug(topic);
-          await ckgUpdateEdge(redis, personId, 'authored', conceptId, outcome === 'success', outcome === 'partial');
-          for (const fp of file_paths.slice(0, 8)) {
-            const fileId = await ckgUpsertFileNode(redis, fp);
-            await ckgUpdateEdge(redis, personId, 'touched', fileId, true);
-          }
-        } catch { /* non-critical */ }
-      }
-
-      // ── Ambient team propagation: if author is set + outcome is success,
-      // auto-store as a team lesson so knowledge is shared without a separate
-      // team_learn call. Tags gain 'team'; the lesson is attributed to the author.
-      // This makes team knowledge sharing the default, not the opt-in.
-      if (author && (outcome === 'success' || outcome === 'partial')) {
-        const teamLesson = {
-          topic, outcome, what_worked,
-          what_failed: what_failed ?? '',
-          severity,
-          author,
-          file_paths,
-          commands,
-          tags: [...new Set([...tags, 'team'])],
-          timestamp: new Date().toISOString(),
-          recall_count: 0,
-          version: 2,
-          auto_propagated: true,
-        };
-        const teamListKey = `cachly:lessons:${topic}`;
-        redis.rpush(teamListKey, JSON.stringify(teamLesson)).catch(() => {});
-      }
-
-      const emoji = outcome === 'success' ? '✅' : outcome === 'partial' ? '⚠️' : '❌';
-      const sevEmoji = severity === 'critical' ? '🔴' : severity === 'major' ? '🟡' : '🟢';
-      const action = isUpdate ? 'updated' : 'stored';
-      return [
-        `${emoji} **Lesson ${action}:** \`${topic}\` (${outcome}) ${sevEmoji} ${severity}`,
-        beliefConflict ?? '',
-        ``,
-        `**What worked:** ${what_worked}`,
-        what_failed ? `**What failed:** ${what_failed}` : '',
-        ctx ? `**Context:** ${ctx}` : '',
-        file_paths.length > 0 ? `**Files:** ${file_paths.map(f => `\`${f}\``).join(', ')}` : '',
-        commands.length > 0 ? `**Commands:** ${commands.map(c => `\`${c}\``).join(', ')}` : '',
-        tags.length > 0 ? `**Tags:** ${tags.map(t => `#${t}`).join(' ')}` : '',
-        ``,
-        isUpdate
-          ? `♻️ Updated (recall count: ${recallCount} · audit entries: ${auditTrail.length})`
-          : `💡 Recall later with \`recall_best_solution(topic="${topic}")\``,
-        (author && (outcome === 'success' || outcome === 'partial'))
-          ? `👥 Auto-shared with team (by _${author}_) — visible in \`team_recall\``
-          : '',
-        depends_on.length > 0
-          ? `🔗 Depends on: ${depends_on.map(d => `\`${d}\``).join(', ')} → trace with \`trace_dependency\``
-          : '',
-        ...templateWarnings,
-        ...iWasWrongWarning,
-      ].filter(l => l !== '').join('\n');
-    }
-
-    case 'recall_best_solution': {
-      const { instance_id, topic } = args as { instance_id: string; topic: string };
-      const redis = await getConnection(instance_id);
-
-      // Try exact best-solution key first
-      const best = await redis.get(`cachly:lesson:best:${topic}`);
-      if (best) {
-        const lesson = safeJsonParse(best, null as null | {
-          topic: string; outcome: string; what_worked: string; what_failed?: string;
-          context?: string; ts: string; verified_at?: string; severity?: string;
-          file_paths?: string[]; commands?: string[]; tags?: string[];
-          recall_count?: number; audit_trail?: unknown[];
-        });
-        if (!lesson) return `⚠️ Lesson data for \`${topic}\` is corrupted. Re-store it with \`learn_from_attempts\`.`;
-
-        // ── Confidence decay check ───────────────────────────────────────────
-        const confidence = calculateConfidence(lesson);
-        const ref = lesson.verified_at ?? lesson.ts;
-        const ageDays = (Date.now() - new Date(ref).getTime()) / 86400000;
-        const badge = confidenceBadge(confidence, ageDays);
-
-        // Recall resets verified_at (confidence clock restart)
-        const updatedLesson = {
-          ...lesson,
-          recall_count: (lesson.recall_count ?? 0) + 1,
-          verified_at: new Date().toISOString(),
-          confidence: 1.0,
-        };
-        await redis.set(`cachly:lesson:best:${topic}`, JSON.stringify(updatedLesson));
-
-        // Track estimated time saved (30m minor · 60m major · 240m critical)
-        const savedMins = lesson.severity === 'critical' ? 240 : lesson.severity === 'major' ? 60 : 30;
-        redis.incrbyfloat(`cachly:stats:time_saved_mins:${instance_id}`, savedMins).catch(() => {});
-
-        const sevEmoji = lesson.severity === 'critical' ? '🔴' : lesson.severity === 'major' ? '🟡' : lesson.severity ? '🟢' : '';
-        const auditSummary = (lesson.audit_trail ?? []).length > 1
-          ? `_Audit: ${(lesson.audit_trail ?? []).length} changes · stored ${new Date(lesson.ts).toLocaleDateString('de-DE')}_`
-          : '';
-
-        // "Remember when..." — emotional header for lessons > 60 days old
-        const ageFromStoreDays = (Date.now() - new Date(lesson.ts).getTime()) / 86400000;
-        const rememberWhen = ageFromStoreDays > 60
-          ? `💭 _Remember when you solved this ${Math.round(ageFromStoreDays / 30)} months ago? Still works._`
-          : '';
-
-        // "Never Google This Again" — suggest pinning after 3rd recall
-        const suggestPin = updatedLesson.recall_count === 3 && !(lesson as { pinned?: boolean }).pinned
-          ? `📌 **You've looked this up 3 times.** Consider pinning it for instant access: add \`pinned: true\` via \`learn_from_attempts\` to always surface it first.`
-          : '';
-
-        // ── Trust signal (today-safe consensus layer) ───────────────────────
-        // A lesson recalled many times — or confirmed by multiple distinct
-        // authors — has proven its value. Surface that as social proof.
-        const lessonAuthors = (lesson as { authors?: string[]; author?: string }).authors
-          ?? ((lesson as { author?: string }).author ? [(lesson as { author?: string }).author!] : []);
-        const distinctAuthors = [...new Set(lessonAuthors.filter(Boolean))];
-        const rc = updatedLesson.recall_count;
-        let trustBadge = '';
-        if (distinctAuthors.length >= 2 && rc >= 5) {
-          trustBadge = `🏆 **Battle-tested** — recalled ${rc}× · verified by ${distinctAuthors.length} developers. Trust this.`;
-        } else if (rc >= 10) {
-          trustBadge = `🏆 **Battle-tested** — recalled ${rc}×. This is one of your most-proven solutions.`;
-        } else if (distinctAuthors.length >= 2) {
-          trustBadge = `✅ **Team-verified** — confirmed by ${distinctAuthors.length} developers.`;
-        } else if (rc >= 5) {
-          trustBadge = `✅ **Proven** — recalled ${rc}× without contradiction.`;
-        }
-
-        return [
-          rememberWhen,
-          trustBadge,
-          `${badge} **Best solution for \`${topic}\`** ${sevEmoji}${lesson.severity ? ` (${lesson.severity})` : ''} · recalled ${updatedLesson.recall_count}×`,
-          ``,
-          `**What worked:** ${lesson.what_worked}`,
-          lesson.what_failed ? `**What failed (avoid this):** ${lesson.what_failed}` : '',
-          lesson.context ? `**Context:** ${lesson.context}` : '',
-          (lesson.file_paths ?? []).length > 0 ? `**Files:** ${(lesson.file_paths ?? []).map((f: string) => `\`${f}\``).join(', ')}` : '',
-          (lesson.commands ?? []).length > 0 ? `**Commands:** ${(lesson.commands ?? []).map((c: string) => `\`${c}\``).join(', ')}` : '',
-          (lesson.tags ?? []).length > 0 ? `**Tags:** ${(lesson.tags ?? []).map((t: string) => `#${t}`).join(' ')}` : '',
-          auditSummary,
-          suggestPin,
-        ].filter(l => l !== '').join('\n');
-      }
-
-      // Partial match: scan all lesson keys for topic substring
-      const allKeys: string[] = [];
-      const scanStream = redis.scanStream({ match: 'cachly:lesson:best:*', count: 100 });
-      await new Promise<void>((resolve, reject) => {
-        scanStream.on('data', (batch: string[]) => allKeys.push(...batch));
-        scanStream.on('end', resolve);
-        scanStream.on('error', reject);
-      });
-
-      const matching = allKeys.filter(k => k.toLowerCase().includes(topic.toLowerCase()));
-      if (matching.length === 0) {
-        // Check attempt history as fallback
-        const histKey = `cachly:lessons:${topic}`;
-        const all = await redis.lrange(histKey, -3, -1);
-        if (all.length > 0) {
-          const parsed = all.map(e => safeJsonParse(e, null as null | { outcome: string; what_worked: string; ts: string })).filter(Boolean) as Array<{ outcome: string; what_worked: string; ts: string }>;
-          const lines = parsed.map(p => `- ${p.outcome === 'success' ? '✅' : '❌'} ${p.what_worked.slice(0, 120)} (${new Date(p.ts).toLocaleDateString('de-DE')})`);
-          return `⚠️ No successful solution for \`${topic}\` yet. Last attempts:\n\n${lines.join('\n')}`;
-        }
-        return `📭 No lessons found for \`${topic}\`. Use \`learn_from_attempts\` after solving it.`;
-      }
-
-      // Return all partial matches
-      const results: string[] = [];
-      for (const k of matching.slice(0, 5)) {
-        const raw = await redis.get(k);
-        if (!raw) continue;
-        const lesson = safeJsonParse(raw, null as null | { topic: string; what_worked: string; context?: string; ts: string });
-        if (!lesson) continue;
-        results.push(`**\`${lesson.topic}\`** — ${lesson.what_worked.slice(0, 200)}`);
-      }
-      return `🔍 **Partial matches for \`${topic}\`:**\n\n${results.join('\n\n')}`;
-    }
-
-    case 'smart_recall': {
-      const {
-        instance_id,
-        query,
-        threshold = 0.78,
-        author: requester = '',
-      } = args as { instance_id: string; query: string; threshold?: number; author?: string };
-
-      const redis = await getConnection(instance_id);
-
-      // ── Layer 1: Keyword search across ALL brain data (always works, no embedding) ──
-      const rawMatches = await keywordSearch(
-        redis,
-        ['cachly:ctx:*', 'cachly:lesson:best:*', 'cachly:idx:*'],
-        query,
-        10,
-      );
-
-      // ── Layer 1.5: Quality-aware rerank — proven success lessons outrank
-      // text-similar failed attempts (the moat; see src/rerank.ts + Cachly-Bench). ──
-      const kwMatches = rerankByQuality(rawMatches);
-
-      // Increment recall_count on matched lessons (fire-and-forget) + collect "Brain saved you here" signal.
-      type RecalledLesson = { topic: string; severity: string; recall_count: number; savedMins: number };
-      const savedHere: RecalledLesson[] = [];
-      const lessonMatches = kwMatches.filter(m => m.key.startsWith('cachly:lesson:best:'));
-      let crossAuthorThisCall = 0;
-      for (const m of lessonMatches.slice(0, 5)) {
-        const existing = await redis.get(m.key).catch(() => null);
-        if (existing) {
-          const lesson = safeJsonParse(existing, null as null | { recall_count?: number; outcome?: string; severity?: string; author?: string; [k: string]: unknown });
-          if (lesson) {
-            const updated = { ...lesson, recall_count: (lesson.recall_count ?? 0) + 1, verified_at: new Date().toISOString() };
-            redis.set(m.key, JSON.stringify(updated)).catch(() => {});
-            const sev = lesson.severity as string;
-            const savedMins = sev === 'critical' ? 240 : sev === 'major' ? 60 : 30;
-            redis.incrbyfloat(`cachly:stats:time_saved_mins:${instance_id}`, savedMins).catch(() => {});
-
-            // Metric 3 (team-knowledge-reuse): a recall of a lesson authored by
-            // someone OTHER than the requester is cross-author reuse — the value
-            // only cachly delivers. Track total proven recalls + cross-author ones.
-            redis.incr(`cachly:stats:recalls_total:${instance_id}`).catch(() => {});
-            if (lesson.author && requester && lesson.author !== requester) {
-              crossAuthorThisCall++;
-              redis.incr(`cachly:stats:cross_author_recalls:${instance_id}`).catch(() => {});
-              redis.sadd(`cachly:stats:reuse_pairs:${instance_id}`, `${requester}<-${lesson.author}`).catch(() => {});
-            }
-
-            // Surface banner for proven successes (recall_count >= 1 means it's been validated)
-            if (lesson.outcome === 'success' && (lesson.recall_count ?? 0) >= 1) {
-              savedHere.push({
-                topic: m.key.replace('cachly:lesson:best:', ''),
-                severity: sev ?? 'major',
-                recall_count: (lesson.recall_count ?? 0) + 1,
-                savedMins,
-              });
-            }
-          }
-        }
-      }
-
-      // Metric 1 (time-to-first-recall): stamp the first time recall returned a
-      // proven lesson. SET NX so only the first successful recall wins.
-      if (savedHere.length > 0) {
-        redis.set(`cachly:stats:first_recall_at:${instance_id}`, new Date().toISOString(), 'EX', 365 * 86400, 'NX').catch(() => {});
-      }
-
-      // ── Layer 2: Semantic search (parallel, optional) ────────────────────────
-      const inst = await apiFetch<Instance | null>(`/api/v1/instances/${instance_id}`).catch(() => null);
-      type SemHit = { key: string; similarity: number; content: string };
-      const semHits: SemHit[] = [];
-      if (inst?.vector_token && hasEmbedProvider()) {
-        try {
-          const embedding = await computeEmbedding(query);
-          const vectorUrl = `https://api.cachly.dev/v1/sem/${inst!.vector_token}`;
-          const searchRes = await fetch(`${vectorUrl}/search`, {
-            method: 'POST',
-            headers: { 'Content-Type': 'application/json' },
-            body: JSON.stringify({ embedding, namespace: 'cachly:ctx', threshold, top_k: 5 }),
-          });
-          if (searchRes.ok) {
-            const results = (await searchRes.json()) as SemanticSearchResponse[];
-            for (const hit of results.filter(r => r.found && r.id)) {
-              const parts = hit.id!.replace('ctx:', '').split(':');
-              const category = parts[0];
-              const key = parts.slice(1).join(':');
-              const redisKey = `cachly:ctx:${category}:${key}`;
-              const content = await redis.get(redisKey).catch(() => null);
-              semHits.push({ key: redisKey, similarity: hit.similarity ?? 0, content: content ?? '(evicted)' });
-            }
-          }
-        } catch {
-          // Semantic search failed silently — keyword results are enough
-        }
-      }
-
-      // ── Layer 3: Hybrid merge — one unified ranked list ──────────────────────
-      // BM25 quality scores (already reranked) are re-normalized to [0,1] for merging.
-      type HybridResult = {
-        key: string; content: string; hybridScore: number;
-        bm25Score?: number; semScore?: number; ckgScore?: number; matchedWords?: string[];
-        matchType: 'keyword' | 'semantic' | 'ckg' | 'hybrid'; subQuery?: string;
-      };
-
-      const bm25Scores = kwMatches.map(m => m.score);
-      const bm25Min = bm25Scores.length ? Math.min(...bm25Scores) : 0;
-      const bm25Range = bm25Scores.length ? (Math.max(...bm25Scores) - bm25Min) || 1 : 1;
-      const bm25Norm = (s: number) => (s - bm25Min) / bm25Range;
-
-      const hybridMap = new Map<string, HybridResult>();
-      for (const m of kwMatches) {
-        const n = bm25Norm(m.score);
-        hybridMap.set(m.key, {
-          key: m.key, content: m.content, bm25Score: n,
-          hybridScore: n * (semHits.length > 0 ? 0.7 : 1.0),
-          matchedWords: m.matchedWords, matchType: 'keyword', subQuery: m.subQuery,
-        });
-      }
-      for (const hit of semHits) {
-        const existing = hybridMap.get(hit.key);
-        if (existing) {
-          existing.semScore = hit.similarity;
-          existing.hybridScore = (existing.bm25Score ?? 0) * 0.6 + hit.similarity * 0.4;
-          existing.matchType = 'hybrid';
-        } else {
-          hybridMap.set(hit.key, {
-            key: hit.key, content: hit.content, semScore: hit.similarity,
-            hybridScore: hit.similarity * 0.5, matchType: 'semantic',
-          });
-        }
-      }
-      // ── Layer 3: CKG traversal — follow "fixes" edges to find lessons that solved
-      // structurally similar problems, even when vocabulary differs ───────────────
-      try {
-        const qTokens = tokenize(query).filter(t => t.length >= 4).slice(0, 4);
-        const candidateNodeIds = new Set<string>();
-        for (const token of qTokens) {
-          const nodeStream = redis.scanStream({ match: `cachly:ckg:node:*${token}*`, count: 20 });
-          await new Promise<void>((res, rej) => {
-            nodeStream.on('data', (batch: string[]) => {
-              for (const k of batch.slice(0, 5)) candidateNodeIds.add(k.replace('cachly:ckg:node:', ''));
-            });
-            nodeStream.on('end', res);
-            nodeStream.on('error', rej);
-          });
-        }
-
-        const seenFroms = new Set<string>();
-        for (const nodeId of [...candidateNodeIds].slice(0, 12)) {
-          const inboundEdgeKeys = await redis.smembers(`cachly:ckg:idx:to:${nodeId}`);
-          for (const ek of inboundEdgeKeys.slice(0, 6)) {
-            const er = await redis.get(ek);
-            if (!er) continue;
-            const edge = safeJsonParse<CKGEdge | null>(er, null);
-            if (!edge || edge.edgeType !== 'fixes' || edge.confidence < 0.35) continue;
-            if (seenFroms.has(edge.from)) continue;
-            seenFroms.add(edge.from);
-
-            const lessonKey = `cachly:lesson:best:${edge.from}`;
-            const existing = hybridMap.get(lessonKey);
-            if (existing) {
-              // Already found by BM25/semantic — boost with causal confirmation
-              existing.ckgScore = Math.max(existing.ckgScore ?? 0, edge.confidence);
-              existing.hybridScore = existing.hybridScore * 0.88 + edge.confidence * 0.12;
-              if (existing.matchType === 'keyword') existing.matchType = 'hybrid';
-            } else {
-              const content = await redis.get(lessonKey).catch(() => null);
-              if (!content) continue;
-              hybridMap.set(lessonKey, {
-                key: lessonKey, content,
-                ckgScore: edge.confidence,
-                hybridScore: edge.confidence * 0.30, // CKG-only: lower than text hits
-                matchType: 'ckg',
-              });
-            }
-          }
-        }
-      } catch { /* non-critical — keyword + semantic always available */ }
-
-      // Filter private lessons — they are only accessible via exact recall_best_solution
-      const hybridResults = [...hybridMap.values()]
-        .filter(r => {
-          if (!r.key.startsWith('cachly:lesson:best:')) return true;
-          const ld = safeJsonParse<{ visibility?: string }>(r.content, {});
-          return ld.visibility !== 'private';
-        })
-        .sort((a, b) => b.hybridScore - a.hybridScore);
-
-      // ── Build output ──────────────────────────────────────────────────────────
-      const lines: string[] = [`🧠 **Smart Recall** for: _"${query}"_\n`];
-
-      // "Brain saved you here" banner — surfaces value inline for proven recalled lessons
-      const SORDER: Record<string, number> = { critical: 0, major: 1, minor: 2 };
-      const topLesson = savedHere.sort((a, b) => (SORDER[a.severity] ?? 1) - (SORDER[b.severity] ?? 1))[0];
-      if (topLesson) {
-        lines.push(`> 💡 **Brain saved you ~${topLesson.savedMins}m here** — ${topLesson.severity} issue from \`${topLesson.topic}\` recalled ${topLesson.recall_count}× proven\n`);
-      }
-      if (crossAuthorThisCall > 0) {
-        lines.push(`> 👥 **Team knowledge reuse** — ${crossAuthorThisCall} of these lesson${crossAuthorThisCall !== 1 ? 's were' : ' was'} written by a teammate. This is the value only a shared brain delivers.\n`);
-      }
-
-      // Show sub-query info if multi-topic was detected
-      const subQueries = splitMultiQuery(query);
-      if (subQueries.length > 1) {
-        lines.push(`_Detected ${subQueries.length} sub-topics:_ ${subQueries.map((s, i) => `${i + 1}. "${s}"`).join(', ')}\n`);
-      }
-
-      if (hybridResults.length > 0) {
-        const hasCKG = hybridResults.some(r => r.ckgScore !== undefined);
-        const modeLabel = semHits.length > 0 && hasCKG
-          ? `keyword + semantic + CKG hybrid`
-          : semHits.length > 0
-          ? `keyword + semantic hybrid`
-          : hasCKG
-          ? `keyword + CKG hybrid`
-          : `keyword`;
-        lines.push(`### 🔍 Results (${hybridResults.length} — ${modeLabel})\n`);
-
-        // Group by sub-query for multi-topic queries
-        if (subQueries.length > 1) {
-          const grouped = new Map<string, HybridResult[]>();
-          for (const r of hybridResults.slice(0, 12)) {
-            const sq = r.subQuery ?? query;
-            if (!grouped.has(sq)) grouped.set(sq, []);
-            grouped.get(sq)!.push(r);
-          }
-          for (const [sq, results] of grouped) {
-            lines.push(`**Topic: "${sq}"** (${results.length} results)\n`);
-            for (const r of results.slice(0, 4)) {
-              const label = r.key.replace('cachly:ctx:', '📝 ').replace('cachly:lesson:best:', '💡 ').replace('cachly:idx:', '📂 ');
-              const scorePart = r.matchType === 'ckg'
-                ? `CKG: ${(r.ckgScore ?? 0).toFixed(2)}, 🕸️ causal graph`
-                : r.ckgScore !== undefined && r.bm25Score !== undefined
-                ? `BM25: ${(r.bm25Score).toFixed(2)}, CKG: ${(r.ckgScore).toFixed(2)}, 🔀 hybrid`
-                : r.matchType === 'hybrid'
-                ? `BM25: ${(r.bm25Score ?? 0).toFixed(2)}, sem: ${((r.semScore ?? 0) * 100).toFixed(0)}%, 🔀 hybrid`
-                : r.matchType === 'semantic'
-                ? `sem: ${((r.semScore ?? 0) * 100).toFixed(0)}%, 🎯 semantic`
-                : `BM25: ${(r.bm25Score ?? 0).toFixed(2)}, matched: ${r.matchedWords?.join(', ')}`;
-              const preview = r.content.slice(0, 300).replace(/\n/g, ' ');
-              lines.push(`  **${label}** _(${scorePart})_`);
-              lines.push(`  > ${preview}${r.content.length > 300 ? '…' : ''}\n`);
-            }
-          }
-          const matched = [...grouped.keys()];
-          const unmatched = subQueries.filter(sq => !matched.includes(sq));
-          if (unmatched.length > 0) {
-            lines.push(`\n⚠️ **No results for:** ${unmatched.map(s => `"${s}"`).join(', ')}`);
-          }
-        } else {
-          for (const r of hybridResults.slice(0, 8)) {
-            const label = r.key.replace('cachly:ctx:', '📝 ').replace('cachly:lesson:best:', '💡 ').replace('cachly:idx:', '📂 ');
-            let authorBadge = '';
-            if (r.key.startsWith('cachly:lesson:best:')) {
-              const ld = safeJsonParse<{ author?: string }>(r.content, {});
-              if (ld.author) authorBadge = ` · 👤 ${ld.author}`;
-            }
-            const scorePart = r.matchType === 'hybrid'
-              ? `BM25: ${(r.bm25Score ?? 0).toFixed(2)}, sem: ${((r.semScore ?? 0) * 100).toFixed(0)}%, 🔀 hybrid`
-              : r.matchType === 'semantic'
-              ? `sem: ${((r.semScore ?? 0) * 100).toFixed(0)}%, 🎯 semantic`
-              : `BM25: ${(r.bm25Score ?? 0).toFixed(2)}, matched: ${r.matchedWords?.join(', ')}`;
-            const preview = r.content.slice(0, 400).replace(/\n/g, ' ');
-            lines.push(`**${label}**${authorBadge} _(${scorePart})_`);
-            lines.push(`> ${preview}${r.content.length > 400 ? '…' : ''}\n`);
-          }
-        }
-      } else {
-        lines.push(`⚠️ No matches found for: "${query}"`);
-
-        // Did-You-Mean: find nearest token in index vocab
-        const queryTokens = tokenize(query);
-        const suggestions: string[] = [];
-        if (_indexVocab.size > 0 && queryTokens.length > 0) {
-          for (const qt of queryTokens.slice(0, 3)) {
-            if (qt.length < 4) continue;
-            let bestDist = 3;
-            let bestTok = '';
-            for (const v of _indexVocab) {
-              if (v.length < 3 || Math.abs(v.length - qt.length) > 4) continue;
-              const d = levenshtein(qt, v);
-              if (d > 0 && d < bestDist) { bestDist = d; bestTok = v; }
-            }
-            if (bestTok) suggestions.push(`"${bestTok}" (instead of "${qt}")`);
-          }
-        }
-        if (suggestions.length > 0) {
-          lines.push(`💡 **Did you mean:** ${suggestions.join(', ')}?`);
-        } else {
-          lines.push(`\n💡 Tips:`);
-          lines.push(`  • Try different keywords`);
-          lines.push(`  • Use \`list_remembered\` to see available context`);
-          lines.push(`  • Use \`recall_best_solution("topic")\` for exact topic lookup`);
-        }
-      }
-
-      return lines.join('\n');
-    }
-
-    // ── get_api_status ────────────────────────────────────────────────────────
-
-    case 'session_start': {
-      const { instance_id, focus = '', author = '', provider = '', workspace_path = '' } = args as { instance_id: string; focus?: string; author?: string; provider?: string; workspace_path?: string };
-      const redis = await getConnection(instance_id);
-
-      // 1. Scan all best-solution lessons
-      const lessonKeys: string[] = [];
-      const lStream = redis.scanStream({ match: 'cachly:lesson:best:*', count: 200 });
-      await new Promise<void>((resolve, reject) => {
-        lStream.on('data', (batch: string[]) => lessonKeys.push(...batch));
-        lStream.on('end', resolve);
-        lStream.on('error', reject);
-      });
-
-      // 2. Fetch all lesson values for recency sorting + focus matching
-      type Lesson = {
-        topic: string; outcome: string; what_worked: string; what_failed?: string;
-        ts: string; verified_at?: string; severity?: string; recall_count?: number;
-        tags?: string[]; confidence?: number; audit_trail?: unknown[];
-      };
-      const lessons: Lesson[] = [];
-      if (lessonKeys.length > 0) {
-        const raws = await redis.mget(...lessonKeys);
-        for (const raw of raws) {
-          const l = safeJsonParse<Lesson | null>(raw ?? null, null);
-          if (l) lessons.push(l);
-        }
-      }
-      lessons.sort((a, b) => new Date(b.ts).getTime() - new Date(a.ts).getTime());
-
-      // 3. Count context entries (filter :meta keys)
-      let ctxCount = 0;
-      const ctxStream = redis.scanStream({ match: 'cachly:ctx:*', count: 200 });
-      await new Promise<void>((resolve, reject) => {
-        ctxStream.on('data', (batch: string[]) => {
-          ctxCount += batch.filter((k: string) => !k.endsWith(':meta')).length;
-        });
-        ctxStream.on('end', resolve);
-        ctxStream.on('error', reject);
-      });
-
-      // 4. Last session
-      const lastSessionRaw = await redis.get('cachly:session:last');
-      type LastSession = { summary: string; ts: string; files_changed?: string[]; duration_min?: number };
-      const lastSession = safeJsonParse<LastSession | null>(lastSessionRaw, null);
-
-      // 5. Focus filtering
-      const focusTerms = focus.toLowerCase().split(/\s+/).filter(Boolean);
-      const focusLessons = focusTerms.length > 0
-        ? lessons.filter(l =>
-            focusTerms.some(term =>
-              l.topic.toLowerCase().includes(term) ||
-              (l.tags ?? []).some((t: string) => t.toLowerCase().includes(term))
-            )
-          )
-        : [];
-
-      // 6. Streak tracking
-      let streakDays = 0;
-      let streakRecord = 0;
-      let streakMessage = '';
-      try {
-        const streakRaw = await redis.get('cachly:streak:current');
-        const streak = safeJsonParse<{ days: number; last_date: string; record: number } | null>(streakRaw, null);
-        const today = new Date().toISOString().slice(0, 10);
-        if (streak) {
-          const lastDate = streak.last_date;
-          const yesterday = new Date(Date.now() - 86400000).toISOString().slice(0, 10);
-          if (lastDate === today) {
-            // Already counted today
-            streakDays = streak.days;
-            streakRecord = streak.record;
-          } else if (lastDate === yesterday) {
-            // Continuing streak
-            streakDays = streak.days + 1;
-            streakRecord = Math.max(streakDays, streak.record);
-            await redis.set('cachly:streak:current', JSON.stringify({ days: streakDays, last_date: today, record: streakRecord }));
-          } else {
-            // Streak broken
-            streakDays = 1;
-            streakRecord = streak.record;
-            await redis.set('cachly:streak:current', JSON.stringify({ days: 1, last_date: today, record: streakRecord }));
-          }
-        } else {
-          // First session ever
-          streakDays = 1;
-          streakRecord = 1;
-          await redis.set('cachly:streak:current', JSON.stringify({ days: 1, last_date: today, record: 1 }));
-        }
-        if (streakDays >= 7) streakMessage = `🔥 **${streakDays}-day streak!** ${streakDays === streakRecord ? ' New record!' : `Best: ${streakRecord}d`}`;
-        else if (streakDays > 1) streakMessage = `🔥 ${streakDays}-day streak`;
-      } catch { /* non-critical */ }
-
-      // 7. Save session start marker
-      await redis.set('cachly:session:current', JSON.stringify({
-        started: new Date().toISOString(),
-        focus,
-        provider,
-      }), 'EX', 86400); // auto-expire after 24h if session_end never called
-
-      // 8. Time saved counter
-      let timeSavedMins = 0;
-      try {
-        const raw = await redis.get(`cachly:stats:time_saved_mins:${instance_id}`);
-        timeSavedMins = parseFloat(raw ?? '0');
-      } catch { /* non-critical */ }
-
-      // ── Build briefing ──────────────────────────────────────────────────────
-      const providerLabel = provider ? ` · ${provider}` : '';
-      const lines: string[] = [`🧠 **Session Briefing**${providerLabel}`, ''];
-      if (streakMessage) lines.push(streakMessage, '');
-
-      // Time saved (only show when meaningful — 30+ minutes)
-      if (timeSavedMins >= 30) {
-        const h = Math.floor(timeSavedMins / 60);
-        const m = Math.round(timeSavedMins % 60);
-        const timeStr = h > 0 ? `${h}h${m > 0 ? ` ${m}m` : ''}` : `${m}m`;
-        lines.push(`⏱️ **Brain saved you ~${timeStr} total** (time not re-researching known fixes)`, '');
-      }
-
-      // ── What's New (shown once per version update) ──────────────────────────
-      try {
-        const seenVersion = await redis.get('cachly:mcp:version:last_seen');
-        if (seenVersion !== MCP_VERSION) {
-          await redis.set('cachly:mcp:version:last_seen', MCP_VERSION, 'EX', 365 * 86400);
-          const changelog = WHATS_NEW[MCP_VERSION];
-          if (changelog) { lines.push(...changelog, ''); }
-        }
-      } catch { /* non-critical */ }
-
-      // ── First-time welcome (empty brain) ───────────────────────────────────
-      const isFirstSession = !lastSession && lessons.length === 0 && ctxCount === 0;
-      if (isFirstSession) {
-        lines.push('🎉 **Welcome! Your AI Brain is live.**', '');
-        if (workspace_path) {
-          lines.push(`🚀 **Auto-bootstrapping from your git history...** (this takes a few seconds)`);
-          lines.push(`   Your brain will learn from your recent commits — no setup needed.`, '');
-        } else {
-          lines.push('It learns from your work automatically. After your first session it will look like this:', '');
-          lines.push('  ✅ `api:auth` — Bearer token in header, not cookie; 401 on missing scope');
-          lines.push('  ✅ `database:migrations` — always run migrations before deploy');
-          lines.push('  ⚠️ `docker:build` — ARG changes bust all subsequent cache layers');
-          lines.push('');
-          lines.push('**Tip:** Pass `workspace_path` to `session_start` to auto-learn from git history instantly.');
-          lines.push('');
-        }
-        lines.push('**Your brain grows automatically:**');
-        lines.push('  • End each session → `session_end(summary="What I did")` — auto-learns from git commits');
-        lines.push('  • After fixing bugs → `learn_from_attempts(topic="...", outcome="success", what_worked="...")`');
-        lines.push('');
-        lines.push('💡 Run `brain_doctor` for a health-check and personalised tips.');
-        lines.push('');
-      }
-
-      // ── Team-virality: first-team-briefing wow moment ────────────────────
-      // When a user joins a team brain (has team lessons from colleagues but
-      // has never been briefed on them), show a dedicated "Welcome to your
-      // team's brain" section. Only fires once per user.
-      if (author && !isFirstSession) {
-        try {
-          const briefingKey = `cachly:team:first_briefing:${author}`;
-          const alreadyBriefed = await redis.get(briefingKey);
-          if (!alreadyBriefed) {
-            type LessonAny = typeof lessons[0] & { author?: string };
-            const teamLessons = (lessons as LessonAny[]).filter(l => l.author && l.author !== author);
-            if (teamLessons.length > 0) {
-              // Mark briefed so this only fires once
-              await redis.set(briefingKey, '1', 'EX', 365 * 86400);
-              const byAuthor = new Map<string, LessonAny[]>();
-              for (const l of teamLessons) {
-                const a = l.author!;
-                if (!byAuthor.has(a)) byAuthor.set(a, []);
-                byAuthor.get(a)!.push(l);
-              }
-              lines.push('');
-              lines.push('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━');
-              lines.push('🤝 **Your team\'s AI brain has been briefing you.**');
-              lines.push('');
-              lines.push(`Your teammates have already solved problems you're about to hit:`);
-              lines.push('');
-              for (const [teamAuthor, tls] of byAuthor) {
-                lines.push(`  👤 **${teamAuthor}** fixed ${tls.length} thing${tls.length > 1 ? 's' : ''}:`);
-                for (const l of tls.slice(0, 2)) {
-                  const emoji = l.outcome === 'success' ? '✅' : '⚠️';
-                  lines.push(`    ${emoji} \`${l.topic}\` — ${l.what_worked.slice(0, 90)}`);
-                }
-                if (tls.length > 2) lines.push(`    … and ${tls.length - 2} more lessons`);
-              }
-              lines.push('');
-              lines.push(`💡 Use \`team_learn\` after your next fix to pay it forward.`);
-              lines.push('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━');
-              lines.push('');
-            }
-          }
-        } catch { /* team briefing errors must never break session_start */ }
-      }
-
-      // Handoff from previous window (if any)
-      const handoffRaw = await redis.get('cachly:session:handoff');
-      if (handoffRaw) {
-        try {
-          const handoff = JSON.parse(handoffRaw) as {
-            ts: string; completed_tasks: string[]; remaining_tasks: string[];
-            files_changed?: { path: string; status: string; description?: string }[];
-            instructions?: string; context_summary?: string; blocked_on?: string;
-          };
-          const ago = Math.round((Date.now() - new Date(handoff.ts).getTime()) / 60000);
-          const agoStr = ago < 60 ? `${ago}m ago` : ago < 1440 ? `${Math.round(ago / 60)}h ago` : `${Math.round(ago / 1440)}d ago`;
-
-          lines.push(`🤝 **Handoff from previous window** (${agoStr}):`);
-          if (handoff.context_summary) lines.push(`   ${handoff.context_summary}`);
-          if (handoff.remaining_tasks.length > 0) {
-            lines.push(`   ⏳ **Remaining tasks:**`);
-            for (const t of handoff.remaining_tasks) lines.push(`     - ${t}`);
-          }
-          if (handoff.completed_tasks.length > 0) {
-            lines.push(`   ✅ **Already done:** ${handoff.completed_tasks.join(', ')}`);
-          }
-          const brokenFiles = (handoff.files_changed ?? []).filter(f => f.status === 'broken' || f.status === 'partial');
-          if (brokenFiles.length > 0) {
-            lines.push(`   ⚠️ **Needs fix:** ${brokenFiles.map(f => `\`${f.path}\` (${f.status}${f.description ? ': ' + f.description : ''})`).join(', ')}`);
-          }
-          if (handoff.blocked_on) lines.push(`   🚫 **Blocked on:** ${handoff.blocked_on}`);
-          if (handoff.instructions) lines.push(`   📝 **Instructions:** ${handoff.instructions}`);
-          lines.push('');
-        } catch { /* ignore corrupt handoff */ }
-      }
-
-      // ── Last checkpoint (session_ping) — shown when no session_end found ────
-      const checkpointRaw = await redis.get('cachly:session:checkpoint');
-      if (checkpointRaw) {
-        try {
-          const cp = JSON.parse(checkpointRaw) as {
-            ts: string; task: string; files_touched: string[]; next_step?: string; provider?: string;
-          };
-          // Only show checkpoint if it's more recent than last session_end
-          const cpTime = new Date(cp.ts).getTime();
-          const lastSessionTime = lastSession ? new Date(lastSession.ts).getTime() : 0;
-          if (cpTime > lastSessionTime) {
-            const ago = Math.round((Date.now() - cpTime) / 60000);
-            const agoStr = ago < 60 ? `${ago}m ago` : ago < 1440 ? `${Math.round(ago / 60)}h ago` : `${Math.round(ago / 1440)}d ago`;
-            const providerStr = cp.provider ? ` [${cp.provider}]` : '';
-            lines.push(`📌 **Last checkpoint**${providerStr} (${agoStr}): ${cp.task}`);
-            if (cp.files_touched.length > 0) {
-              lines.push(`   Files: ${cp.files_touched.slice(0, 5).map(f => `\`${f}\``).join(', ')}`);
-            }
-            if (cp.next_step) lines.push(`   📍 Next step was: ${cp.next_step}`);
-            if (!lastSession || cpTime - lastSessionTime > 300_000) {
-              lines.push(`   ⚠️ No \`session_end\` found — reconstructed from last checkpoint`);
-            }
-            lines.push('');
-          }
-        } catch { /* ignore */ }
-      }
-
-      // ── Git reconstruction — when no session_end + workspace_path given ─────
-      if (workspace_path && !lastSession) {
-        try {
-          const { execSync } = await import('node:child_process');
-          const gitLog = execSync(
-            `git -C "${workspace_path}" log --oneline --format="%h %s" -15 2>/dev/null`,
-            { encoding: 'utf-8', timeout: 5000 },
-          ).trim();
-          const gitDiff = execSync(
-            `git -C "${workspace_path}" diff --stat HEAD~3 2>/dev/null || git -C "${workspace_path}" diff --stat 2>/dev/null`,
-            { encoding: 'utf-8', timeout: 5000 },
-          ).trim();
-          if (gitLog) {
-            lines.push(`🔍 **Git reconstruction** (no session_end found — reconstructed from git):`);
-            for (const l of gitLog.split('\n').slice(0, 8)) lines.push(`   ${l}`);
-            if (gitDiff) {
-              const diffLines = gitDiff.split('\n').filter(l => l.includes('|') || l.includes('changed'));
-              if (diffLines.length > 0) {
-                lines.push(`   **Recent changes:**`);
-                for (const dl of diffLines.slice(0, 5)) lines.push(`   ${dl.trim()}`);
-              }
-            }
-            lines.push('');
-          }
-        } catch { /* git not available or no repo — silent */ }
-      }
-
-      // Last session
-      if (lastSession) {
-        const ago = Math.round((Date.now() - new Date(lastSession.ts).getTime()) / 60000);
-        const agoStr = ago < 60 ? `${ago}m ago` : ago < 1440 ? `${Math.round(ago / 60)}h ago` : `${Math.round(ago / 1440)}d ago`;
-        lines.push(`📅 **Last session** (${agoStr}): ${lastSession.summary}`);
-        if (lastSession.duration_min) lines.push(`   Duration: ${lastSession.duration_min} min`);
-        if ((lastSession.files_changed ?? []).length > 0) {
-          lines.push(`   Files: ${(lastSession.files_changed ?? []).slice(0, 5).map((f: string) => `\`${f}\``).join(', ')}`);
-        }
-        lines.push('');
-      }
-
-      // Brain health
-      lines.push(`📊 **Brain:** ${lessons.length} lessons · ${ctxCount} context entries`, '');
-
-      // ── 📌 Proven Laws — auto-crystallized lessons (quick win 3) ─────────────
-      // A lesson that has proven itself (recalled ≥ CRYSTALLIZE_RECALLS times) or
-      // was explicitly pinned becomes a "law" — surfaced at the top of every
-      // briefing so the most battle-tested knowledge is never buried. This is
-      // the today-safe version of swarm crystallization: enough confirmations
-      // auto-promote a lesson without any manual step.
-      {
-        type LessonPin = typeof lessons[0] & { pinned?: boolean };
-        const CRYSTALLIZE_RECALLS = 5;
-        const laws = (lessons as LessonPin[])
-          .filter(l => l.outcome === 'success' && (l.pinned === true || (l.recall_count ?? 0) >= CRYSTALLIZE_RECALLS))
-          .sort((a, b) => (b.recall_count ?? 0) - (a.recall_count ?? 0))
-          .slice(0, 5);
-        if (laws.length > 0) {
-          lines.push(`📌 **Proven Laws** (auto-crystallized — your most-trusted knowledge):`);
-          for (const l of laws) {
-            const sev = l.severity === 'critical' ? '🔴' : l.severity === 'major' ? '🟡' : '';
-            const rc  = l.recall_count ?? 0;
-            const why = l.pinned === true ? 'pinned' : `recalled ${rc}×`;
-            lines.push(`  🏆${sev} \`${l.topic}\` _(${why})_ — ${l.what_worked.slice(0, 100)}`);
-          }
-          lines.push('');
-        }
-      }
-
-      // ── Layer 7: MCM Domain Coverage Map ────────────────────────────────────
-      if (lessons.length >= 3) {
-        const domainMap = new Map<string, { total: number; success: number; critical: number }>();
-        for (const l of lessons) {
-          const dom = l.topic.split(':')[0] ?? 'other';
-          if (!domainMap.has(dom)) domainMap.set(dom, { total: 0, success: 0, critical: 0 });
-          const d = domainMap.get(dom)!;
-          d.total++;
-          if (l.outcome === 'success') d.success++;
-          if (l.severity === 'critical') d.critical++;
-        }
-        const sorted = [...domainMap.entries()].sort((a, b) => b[1].total - a[1].total).slice(0, 6);
-        const hasContestedDomains = sorted.some(([, d]) => d.success < d.total * 0.4 && d.total >= 2);
-        if (sorted.length > 0) {
-          lines.push(`🗺️ **Knowledge Coverage:**`);
-          for (const [dom, d] of sorted) {
-            const pct = Math.round((d.success / d.total) * 100);
-            const filled = Math.round(pct / 10);
-            const bar = '█'.repeat(filled) + '░'.repeat(10 - filled);
-            const flag = d.critical > 0 ? ' 🔴' : pct < 40 && d.total >= 2 ? ' ⚠️' : '';
-            lines.push(`  ${bar} ${dom.padEnd(18)} ${String(pct).padStart(3)}% (${d.success}/${d.total} confirmed)${flag}`);
-          }
-          if (hasContestedDomains) {
-            lines.push(`  ⚠️ _Some domains have contested beliefs — use \`ckg_inspect\` to review_`);
-          }
-          lines.push('');
-        }
-      }
-
-      // ── Layer 7 MCM: Active belief conflicts ─────────────────────────────────
-      try {
-        const conflictKeys: string[] = [];
-        const cfStream = redis.scanStream({ match: 'cachly:ckg:conflict:*', count: 50 });
-        await new Promise<void>((res, rej) => { cfStream.on('data', (b: string[]) => conflictKeys.push(...b)); cfStream.on('end', res); cfStream.on('error', rej); });
-        if (conflictKeys.length > 0) {
-          lines.push(`⚡ **Active belief conflicts (${conflictKeys.length}):**`);
-          for (const ck of conflictKeys.slice(0, 3)) {
-            const cr = await redis.get(ck);
-            if (!cr) continue;
-            const cf = safeJsonParse<{ topic: string; fix_confidence: number; fix_trials: number } | null>(cr, null);
-            if (!cf) continue;
-            lines.push(`  ⚠️ \`${cf.topic}\` — previously confirmed fix (${(cf.fix_confidence * 100).toFixed(0)}%, n=${cf.fix_trials}) now contradicted. Use \`ckg_inspect(concept="${ckgSlug(cf.topic)}")\``);
-          }
-          lines.push('');
-        }
-      } catch { /* non-critical */ }
-
-      // ── Layer 7 MCM: Blind Spot Detection ────────────────────────────────────
-      // If the focus mentions a domain that has no CKG node → surface blind spot
-      if (focus && focus.length > 3) {
-        try {
-          const focusTokens = focus.toLowerCase().replace(/[^a-z0-9\s:_-]/g, ' ').split(/\s+/).filter(t => t.length > 3);
-          const blindSpots: string[] = [];
-          for (const token of focusTokens.slice(0, 6)) {
-            const nodeExists = await redis.exists(`cachly:ckg:node:${token}`);
-            if (!nodeExists) {
-              // Check if any node starts with this token (prefix match)
-              const prefixKeys: string[] = [];
-              const psStream = redis.scanStream({ match: `cachly:ckg:node:${token}*`, count: 10 });
-              await new Promise<void>((res, rej) => { psStream.on('data', (b: string[]) => prefixKeys.push(...b)); psStream.on('end', res); psStream.on('error', rej); });
-              if (prefixKeys.length === 0) blindSpots.push(token);
-            }
-          }
-          if (blindSpots.length > 0) {
-            lines.push(`🔭 **Blind spots detected for this focus:**`);
-            for (const bs of blindSpots.slice(0, 3)) {
-              lines.push(`  ⬜ \`${bs}\` — no CKG knowledge. Suggestions:`);
-              lines.push(`     • \`brain_from_git(instance_id="...", concept="${bs}")\` — bootstrap from commits`);
-              lines.push(`     • \`fedbrain_search(query="${bs}")\` — search global commons`);
-            }
-            lines.push('');
-          }
-        } catch { /* non-critical */ }
-      }
-
-      if (focusLessons.length > 0) {
-        lines.push(`🎯 **Relevant for "${focus}":**`);
-        for (const l of focusLessons.slice(0, 4)) {
-          const emoji = l.outcome === 'success' ? '✅' : l.outcome === 'partial' ? '⚠️' : '❌';
-          const sev = l.severity === 'critical' ? '🔴' : l.severity === 'major' ? '🟡' : '';
-          lines.push(`  ${emoji}${sev} \`${l.topic}\` — ${l.what_worked.slice(0, 100)}`);
-        }
-        lines.push('');
-      }
-
-      // Top lessons — sorted by recall_count desc (most-used = most proven value).
-      // If focus is set, focus-matched lessons already shown above; show remaining here.
-      if (lessons.length > 0) {
-        const byRecall = [...lessons].sort((a, b) => (b.recall_count ?? 0) - (a.recall_count ?? 0));
-        const topRecalled = byRecall.filter(l => (l.recall_count ?? 0) > 0).slice(0, 3);
-        const remaining   = (focusLessons.length > 0 ? lessons.filter(l => !focusLessons.includes(l)) : lessons)
-          .filter(l => !topRecalled.includes(l))
-          .slice(0, focusLessons.length > 0 ? 3 : 4);
-
-        if (topRecalled.length > 0) {
-          lines.push(`🏆 **Most valuable** (recalled ${topRecalled[0]!.recall_count}× before):`);
-          for (const l of topRecalled) {
-            const emoji = l.outcome === 'success' ? '✅' : l.outcome === 'partial' ? '⚠️' : '❌';
-            const sev   = l.severity === 'critical' ? '🔴' : l.severity === 'major' ? '🟡' : '';
-            const rc    = l.recall_count ?? 0;
-            const la     = (l as { authors?: string[]; author?: string }).authors
-              ?? ((l as { author?: string }).author ? [(l as { author?: string }).author!] : []);
-            const nAuthors = new Set(la.filter(Boolean)).size;
-            const trust  = (nAuthors >= 2 && rc >= 5) || rc >= 10 ? ' 🏆'
-              : nAuthors >= 2 || rc >= 5 ? ' ✅' : '';
-            const rct   = rc > 1 ? ` _(${rc}× recalled)_` : '';
-            lines.push(`  ${emoji}${sev}${trust} \`${l.topic}\`${rct} — ${l.what_worked.slice(0, 100)}`);
-          }
-          lines.push('');
-        }
-
-        if (remaining.length > 0) {
-          const header = topRecalled.length > 0 ? `🕐 **Recent:**` : `🕐 **Recent lessons:**`;
-          lines.push(header);
-          for (const l of remaining) {
-            const emoji = l.outcome === 'success' ? '✅' : l.outcome === 'partial' ? '⚠️' : '❌';
-            const sev   = l.severity === 'critical' ? '🔴' : l.severity === 'major' ? '🟡' : '';
-            lines.push(`  ${emoji}${sev} \`${l.topic}\` — ${l.what_worked.slice(0, 100)}`);
-          }
-          lines.push('');
-        }
-      } else {
-        lines.push('📭 No lessons yet. Use `learn_from_attempts` after solving tasks.', '');
-      }
-
-      // Team invite prompt — fires once after 10th lesson if no team use yet
-      if (lessons.length >= 10) {
-        const hasTeamUse = lessons.some(l => (l as typeof l & { author?: string }).author);
-        if (!hasTeamUse) {
-          lines.push(`🤝 **Your Brain has ${lessons.length} lessons — your team could benefit instantly.**`);
-          lines.push(`   Share: **cachly.dev/team** · or run \`team_learn\` after the next fix.`, '');
-        }
-      }
-
-      // Open failures (lessons whose best-key has outcome != success)
-      const openFailures = lessons.filter(l => l.outcome === 'failure' || l.outcome === 'partial');
-      if (openFailures.length > 0) {
-        lines.push(`⚠️ **Unresolved** (${openFailures.length} topic${openFailures.length > 1 ? 's' : ''} with no success yet):`);
-        for (const l of openFailures.slice(0, 3)) {
-          lines.push(`  ❌ \`${l.topic}\` — ${(l.what_failed ?? l.what_worked).slice(0, 80)}`);
-        }
-        lines.push('');
-      }
-
-      // ── Stale / low-confidence lessons (confidence decay) ─────────────────
-      const staleSuccessLessons = lessons.filter(l => {
-        if (l.outcome !== 'success' && l.outcome !== 'partial') return false;
-        return calculateConfidence(l) < CONFIDENCE_WARN_VALUE;
-      });
-      if (staleSuccessLessons.length > 0) {
-        lines.push(`🔴 **Stale lessons** (not recalled in >${CONFIDENCE_WARN_DAYS}d — verify before applying):`);
-        for (const l of staleSuccessLessons.slice(0, 4)) {
-          const conf = calculateConfidence(l);
-          const ageDays = Math.round((Date.now() - new Date(l.verified_at ?? l.ts).getTime()) / 86400000);
-          const flag = conf < CONFIDENCE_STALE_VALUE ? '🔴' : '⚠️';
-          lines.push(`  ${flag} \`${l.topic}\` — ${ageDays}d stale, ${(conf * 100).toFixed(0)}% confidence`);
-        }
-        lines.push(`  _Run \`recall_best_solution\` on these to reset their confidence clock._`);
-        lines.push('');
-      }
-
-      // ── Session Replay: show last session's decision log ──────────────────
-      const lastSessionAny = lastSession as unknown as { decision_log?: Array<{ topic: string; outcome: string; what_worked: string }> } | null;
-      if (lastSessionAny?.decision_log?.length) {
-        const dl = lastSessionAny.decision_log;
-        const successes = dl.filter(d => d.outcome === 'success');
-        const failures  = dl.filter(d => d.outcome === 'failure');
-        lines.push(`🎬 **Last session decisions** (${dl.length} lessons stored):`);
-        if (successes.length > 0) lines.push(`  ✅ Worked: ${successes.slice(0, 3).map(d => `\`${d.topic}\``).join(', ')}`);
-        if (failures.length > 0)  lines.push(`  ❌ Failed: ${failures.slice(0, 3).map(d => `\`${d.topic}\``).join(', ')}`);
-        lines.push('');
-      }
-
-      // ── 🔮 Predictive Pre-Warning — intent-based danger detection ────────────
-      // Fires BEFORE work starts. Uses explicit focus when given; otherwise
-      // derives the likely work area from the last session's changed files +
-      // summary (you usually keep working where you left off). This makes the
-      // warning fire even when the caller forgets to pass `focus`.
-      {
-        type LessonAny = typeof lessons[0] & { author?: string; tags?: string[] };
-
-        // Common path noise / extensions to drop when deriving terms from files.
-        const PATH_NOISE = new Set([
-          'src', 'lib', 'test', 'tests', 'dist', 'index', 'main', 'app',
-          'internal', 'pkg', 'cmd', 'node_modules', 'components', 'utils',
-          'ts', 'tsx', 'js', 'jsx', 'go', 'py', 'rs', 'java', 'json', 'yaml', 'yml',
-        ]);
-        const deriveTermsFromFiles = (files: string[]): string[] => {
-          const terms = new Set<string>();
-          for (const f of files) {
-            for (const seg of f.toLowerCase().split(/[/\\._-]/)) {
-              if (seg.length > 3 && !PATH_NOISE.has(seg)) terms.add(seg);
-            }
-          }
-          return [...terms];
-        };
-
-        let warnTerms = focusTerms;
-        let warnLabel = focus;
-        let derived = false;
-        if (warnTerms.length === 0 && lastSession) {
-          const fromFiles = deriveTermsFromFiles(lastSession.files_changed ?? []);
-          const fromSummary = (lastSession.summary ?? '').toLowerCase()
-            .replace(/[^a-z0-9\s:_-]/g, ' ').split(/\s+/).filter(t => t.length > 3 && !PATH_NOISE.has(t));
-          warnTerms = [...new Set([...fromFiles, ...fromSummary])];
-          warnLabel = 'where you left off last session';
-          derived = true;
-        }
-
-        if (warnTerms.length > 0) {
-          const dangerLessons = (lessons as LessonAny[]).filter(l => {
-            if (l.outcome === 'success') return false;
-            const topicCategory = l.topic.split(':')[0];
-            return warnTerms.some(term =>
-              l.topic.toLowerCase().includes(term) ||
-              topicCategory === term ||
-              (l.tags ?? []).some((t: string) => t.toLowerCase() === term),
-            );
-          });
-          if (dangerLessons.length >= 1) {
-            const headline = derived
-              ? `  You're likely to continue **${warnLabel}** — ${dangerLessons.length} known pitfall${dangerLessons.length > 1 ? 's' : ''} there:`
-              : `  Known pitfalls for **"${warnLabel}"** (${dangerLessons.length} past failure${dangerLessons.length > 1 ? 's' : ''}):`;
-            const warning = [
-              `🚨 **PRE-WARNING** — Read this BEFORE starting:`,
-              headline,
-              ...dangerLessons.slice(0, 3).map(l => `  ❌ \`${l.topic}\` — ${(l.what_failed ?? l.what_worked).slice(0, 80)}`),
-              '',
-            ];
-            lines.splice(2, 0, ...warning); // after '🧠 **Session Briefing**' + empty line
-          }
-        }
-      }
-
-      // ── 👥 Team Telepathy — what teammates learned this week ─────────────────
-      if (author) {
-        type LessonAny = typeof lessons[0] & { author?: string };
-        const oneWeekAgo = Date.now() - 7 * 86_400_000;
-        const teamLessons = (lessons as LessonAny[]).filter(l =>
-          l.author && l.author !== author && new Date(l.ts).getTime() > oneWeekAgo,
-        );
-        if (teamLessons.length > 0) {
-          // Group by author
-          const byAuthor = new Map<string, LessonAny[]>();
-          for (const l of teamLessons) {
-            const a = l.author!;
-            if (!byAuthor.has(a)) byAuthor.set(a, []);
-            byAuthor.get(a)!.push(l);
-          }
-          lines.push(`👥 **Team this week** (${teamLessons.length} lesson${teamLessons.length > 1 ? 's' : ''} from teammates):`);
-          for (const [teamAuthor, tls] of byAuthor) {
-            lines.push(`  👤 **${teamAuthor}**:`);
-            for (const l of tls.slice(0, 3)) {
-              const emoji = l.outcome === 'success' ? '✅' : l.outcome === 'partial' ? '⚠️' : '❌';
-              lines.push(`    ${emoji} \`${l.topic}\` — ${l.what_worked.slice(0, 80)}`);
-            }
-            if (tls.length > 3) lines.push(`    … and ${tls.length - 3} more`);
-          }
-          lines.push('');
-        }
-      }
-
-      // ── 💎 Memory Crystal — compressed wisdom from old sessions ──────────────
-      try {
-        const crystalRaw = await redis.get('cachly:crystal:latest');
-        if (crystalRaw) {
-          type CrystalData = { label: string; ts: string; session_count: number; top_patterns: Array<{ category: string; insight: string; count: number }> };
-          const crystal = safeJsonParse<CrystalData | null>(crystalRaw, null);
-          if (crystal) {
-            const crystalAge = Math.round((Date.now() - new Date(crystal.ts).getTime()) / 86_400_000);
-            if (crystalAge <= 90) {
-              lines.push(`💎 **Memory Crystal** (${crystal.label} · ${crystal.session_count} sessions compressed):`);
-              for (const p of crystal.top_patterns.slice(0, 3)) {
-                lines.push(`  • **${p.category}** (${p.count}×): ${p.insight.slice(0, 90)}`);
-              }
-              lines.push('');
-            }
-          }
-        }
-      } catch { /* non-critical */ }
-
-      // ── 🗺️ Roadmap — open items at session start ────────────────────────────
-      try {
-        const roadmapAll = await redis.hgetall(`cachly:roadmap:${instance_id}`);
-        if (roadmapAll && Object.keys(roadmapAll).length > 0) {
-          const PRIORITY_ORDER: Record<string, number> = { critical: 0, high: 1, medium: 2, low: 3 };
-          const PRIORITY_ICON: Record<string, string> = { critical: '🔴', high: '🟠', medium: '🟡', low: '🔵' };
-          const openStatuses = new Set(['planned', 'in-progress', 'blocked']);
-          const allItems = Object.values(roadmapAll).flatMap(v => {
-            const item = safeJsonParse<Record<string, unknown> | null>(v as string, null);
-            return item ? [item] : [];
-          });
-          const openItems = allItems
-            .filter(i => openStatuses.has(i.status as string))
-            .sort((a, b) => {
-              if (a.status === 'in-progress' && b.status !== 'in-progress') return -1;
-              if (b.status === 'in-progress' && a.status !== 'in-progress') return 1;
-              return (PRIORITY_ORDER[a.priority as string] ?? 99) - (PRIORITY_ORDER[b.priority as string] ?? 99);
-            });
-          const doneCount = allItems.filter(i => i.status === 'done').length;
-          if (openItems.length > 0) {
-            lines.push(`🗺️ **Roadmap** (${openItems.length} open · ${doneCount} done):`);
-            for (const it of openItems.slice(0, 5)) {
-              const statusIcon = it.status === 'in-progress' ? '⚡' : it.status === 'blocked' ? '🚫' : '📋';
-              lines.push(`  ${statusIcon} ${PRIORITY_ICON[it.priority as string] ?? '⚪'} \`${it.id}\` **${it.title}**`);
-            }
-            if (openItems.length > 5) lines.push(`  … and ${openItems.length - 5} more`);
-            lines.push(`  _Use \`roadmap_next\` for the top priority item · \`roadmap_list\` for full view_`);
-            lines.push('');
-          }
-        }
-      } catch { /* non-critical */ }
-
-      // ── 🔮 PPE: Predictive Pre-fetch — CKG-powered risk detection ────────────
-      // Layer 4: Before starting any work, scan the CKG for nodes matching focus
-      // tokens and traverse causal edges to surface predicted failure points inline.
-      // This is the PPE "pre-fetch" that was previously only available via brain_predict.
-      if (focus && focus.length > 3) {
-        try {
-          const ppeFocusTokens = focus.toLowerCase().replace(/[^a-z0-9\s\-_:]/g, ' ').split(/\s+/).filter(t => t.length > 2).slice(0, 5);
-          type PPEPrediction = { concept: string; edgeType: string; target: string; confidence: number; lesson?: { what_worked?: string; topic: string } };
-          const ppePredictions: PPEPrediction[] = [];
-
-          for (const token of ppeFocusTokens) {
-            const nodeKeys: string[] = [];
-            const nStream = redis.scanStream({ match: `cachly:ckg:node:*${token}*`, count: 20 });
-            await new Promise<void>((res, rej) => { nStream.on('data', (b: string[]) => nodeKeys.push(...b)); nStream.on('end', res); nStream.on('error', rej); });
-
-            for (const nk of nodeKeys.slice(0, 3)) {
-              const nodeRaw = await redis.get(nk);
-              if (!nodeRaw) continue;
-              const node = safeJsonParse<CKGNode | null>(nodeRaw, null);
-              if (!node) continue;
-              const edgeKeys = await redis.smembers(`cachly:ckg:idx:from:${node.id}`);
-              for (const ek of edgeKeys.slice(0, 15)) {
-                const edgeRaw = await redis.get(ek);
-                if (!edgeRaw) continue;
-                const edge = safeJsonParse<CKGEdge | null>(edgeRaw, null);
-                if (!edge) continue;
-                if (edge.edgeType !== 'causes' && edge.edgeType !== 'co-occurs' && edge.edgeType !== 'fixes') continue;
-                if (edge.confidence < 0.35) continue;
-                const lessonKey = edge.edgeType === 'fixes' ? `cachly:lesson:best:${edge.from}` : `cachly:lesson:best:${edge.to}`;
-                const lessonRaw = await redis.get(lessonKey);
-                const lesson = safeJsonParse<{ what_worked?: string; topic: string } | null>(lessonRaw, null);
-                ppePredictions.push({ concept: node.id, edgeType: edge.edgeType, target: edge.to, confidence: edge.confidence, lesson: lesson ?? undefined });
-              }
-            }
-          }
-
-          const ppeSeen = new Set<string>();
-          const ppeUniq = ppePredictions
-            .filter(p => { const k = `${p.concept}:${p.target}`; if (ppeSeen.has(k)) return false; ppeSeen.add(k); return true; })
-            .sort((a, b) => b.confidence - a.confidence)
-            .slice(0, 4);
-
-          if (ppeUniq.length > 0) {
-            lines.push(`🔮 **Predicted risks for "${focus}"** (PPE pre-fetch):`);
-            for (const p of ppeUniq) {
-              const confPct = Math.round(p.confidence * 100);
-              const icon = p.edgeType === 'causes' ? '⚡' : p.edgeType === 'fixes' ? '🔧' : '🔄';
-              lines.push(`  ${icon} **${confPct}%** \`${p.concept}\` ${p.edgeType} \`${p.target}\``);
-              if (p.lesson?.what_worked) lines.push(`     ✅ ${(p.lesson.what_worked as string).slice(0, 110)}`);
-            }
-            lines.push(`  _Full analysis: \`brain_predict(context="${focus}")\`_`);
-            lines.push('');
-          }
-        } catch { /* non-critical — never block session start */ }
-      }
-
-      // ── 📐 MCM Confidence Calibration — 30-day accuracy pass ─────────────────
-      // Layer 7: Periodically check if high-confidence lessons are actually reliable.
-      // If last calibration was >30 days ago (or never), run a quick pass.
-      // Measures: of recalled lessons with confidence > 0.85, what % had outcome=success?
-      try {
-        const calRaw = await redis.get('cachly:mcm:calibration:last');
-        const lastCalMs = safeJsonParse<{ ts?: number } | null>(calRaw, null)?.ts ?? 0;
-        const daysSinceCal = (Date.now() - lastCalMs) / 86_400_000;
-        if (daysSinceCal >= 30 && lessons.length >= 5) {
-          // Quick calibration pass on recalled lessons (recall_count > 0, outcome=success)
-          const recalledSuccess = lessons.filter(l => l.outcome === 'success' && (l.recall_count ?? 0) > 0);
-          const recalledAll     = lessons.filter(l => (l.recall_count ?? 0) > 0);
-          if (recalledAll.length >= 3) {
-            const precision = recalledAll.length > 0 ? recalledSuccess.length / recalledAll.length : 1;
-            const lowPrecisionDomains: string[] = [];
-            // Per-domain breakdown
-            const domCalMap = new Map<string, { success: number; total: number }>();
-            for (const l of recalledAll) {
-              const dom = l.topic.split(':')[0] ?? 'other';
-              if (!domCalMap.has(dom)) domCalMap.set(dom, { success: 0, total: 0 });
-              const d = domCalMap.get(dom)!;
-              d.total++;
-              if (l.outcome === 'success') d.success++;
-            }
-            for (const [dom, d] of domCalMap) {
-              if (d.total >= 2 && d.success / d.total < 0.6) lowPrecisionDomains.push(dom);
-            }
-            // Save calibration result
-            await redis.set('cachly:mcm:calibration:last', JSON.stringify({
-              ts: Date.now(),
-              precision: precision,
-              recalled: recalledAll.length,
-              low_precision_domains: lowPrecisionDomains,
-            }));
-            if (lowPrecisionDomains.length > 0 || precision < 0.7) {
-              lines.push(`📐 **MCM Calibration** (30-day pass — ${recalledAll.length} recalled lessons):`);
-              lines.push(`  Overall precision: **${(precision * 100).toFixed(0)}%** recalled lessons actually worked`);
-              if (lowPrecisionDomains.length > 0) {
-                lines.push(`  ⚠️ Low-precision domains: ${lowPrecisionDomains.map(d => `\`${d}\``).join(', ')} — consider revisiting these lessons`);
-              }
-              lines.push(`  💡 Use \`learn_from_attempts\` to update stale lessons and improve accuracy.`);
-              lines.push('');
-            }
-          }
-          // Even if no issues, stamp the date so we don't re-check for 30d
-          await redis.set('cachly:mcm:calibration:last', JSON.stringify({ ts: Date.now(), precision: 1, recalled: 0 }), 'EX', 35 * 86400);
-        }
-      } catch { /* non-critical */ }
-
-      // ── 🌍 Knowledge Commons — community stats banner ───────────────────────
-      try {
-        const commonsStats = await apiFetch<{
-          total_lessons: number;
-          total_confirms: number;
-          added_last_7_days: number;
-        }>('/api/v1/syndication/stats');
-        if (commonsStats.total_lessons > 0) {
-          lines.push(
-            `🌍 **Commons:** ${commonsStats.total_lessons.toLocaleString()} lessons · ` +
-            `${commonsStats.total_confirms.toLocaleString()} confirms · ` +
-            `+${commonsStats.added_last_7_days} this week`,
-          );
-          lines.push('');
-        }
-      } catch { /* non-critical — never block session start */ }
-
-      // ── Brain Doctor hint (Punkt 6) — surface when brain needs attention ───
-      const hasOpenFailures = lessons.filter(l => l.outcome === 'failure' || l.outcome === 'partial').length > 0;
-      const hasStaleLessons = lessons.some(l => l.outcome === 'success' && calculateConfidence(l) < CONFIDENCE_WARN_VALUE);
-      if (lessons.length === 0 || hasOpenFailures || hasStaleLessons) {
-        const reasons: string[] = [];
-        if (lessons.length === 0) reasons.push('brain is empty');
-        if (hasOpenFailures) reasons.push('open failures');
-        if (hasStaleLessons) reasons.push('stale lessons');
-        lines.push(`🩺 _Run \`brain_doctor\` to fix: ${reasons.join(', ')}._`);
-      }
-
-      return lines.join('\n');
-    }
-
-    // ── session_end ───────────────────────────────────────────────────────────
-    case 'session_end': {
-      const {
-        instance_id,
-        summary,
-        files_changed = [],
-        lessons_learned,
-        workspace_path = '',
-      } = args as {
-        instance_id: string;
-        summary: string;
-        files_changed?: string[];
-        lessons_learned?: number;
-        workspace_path?: string;
-      };
-
-      const redis = await getConnection(instance_id);
-      const now = new Date();
-
-      // Calculate duration from session_start marker
-      let durationMin: number | undefined;
-      const currentRaw = await redis.get('cachly:session:current');
-      if (!currentRaw) {
-        return `⚠️ No active session found.\n\nRun \`session_start\` first to begin tracking a session, then call \`session_end\` when you're done.`;
-      }
-      const current = safeJsonParse<{ started: string } | null>(currentRaw, null);
-      if (current?.started) {
-        durationMin = Math.round((now.getTime() - new Date(current.started).getTime()) / 60000);
-      }
-
-      // ── Session Replay: capture decision log ─────────────────────────────
-      type DecisionEntry = { ts: string; topic: string; outcome: string; what_worked: string };
-      let decisionLog: DecisionEntry[] = [];
-      try {
-        const dlEntries = await redis.lrange('cachly:session:decision-log', 0, -1);
-        decisionLog = dlEntries.flatMap(e => {
-          const entry = safeJsonParse<DecisionEntry | null>(e, null);
-          return entry ? [entry] : [];
-        });
-        await redis.del('cachly:session:decision-log');
-      } catch { /* non-critical */ }
-
-      const sessionRecord = {
-        ts: now.toISOString(),
-        summary,
-        files_changed,
-        ...(lessons_learned !== undefined ? { lessons_learned } : {}),
-        ...(durationMin !== undefined ? { duration_min: durationMin } : {}),
-        ...(decisionLog.length > 0 ? { decision_log: decisionLog } : {}),
-      };
-
-      // Save as "last session"
-      await redis.set('cachly:session:last', JSON.stringify(sessionRecord));
-
-      // Append to history list (keep last 50 sessions, TTL 90 days)
-      await redis.lpush('cachly:session:history', JSON.stringify(sessionRecord));
-      await redis.ltrim('cachly:session:history', 0, 49);
-      await redis.expire('cachly:session:history', 90 * 86400);
-
-      // Clean up current session marker
-      await redis.del('cachly:session:current');
-
-      // ── AUTO-LEARN from session summary (no manual call needed) ─────────────
-      // Parse the summary for actionable lessons and store them automatically.
-      const autoLearned: string[] = [];
-      try {
-        // Extract key sentences from the summary that contain action verbs
-        const actionVerbs = /\b(fixed|deployed|added|removed|refactored|migrated|updated|resolved|implemented|improved|optimized|configured|created|deleted|disabled|enabled|discovered|found|learned|debugged|patched|upgraded|installed|tested|built|rewrote|moved|renamed|split|merged|extracted)\b/i;
-        const sentences = summary
-          .split(/[.!\n]+/)
-          .map(s => s.trim())
-          .filter(s => s.length > 20 && actionVerbs.test(s));
-
-        for (const sentence of sentences.slice(0, 6)) {
-          // Build a topic slug from the first meaningful words
-          const words = sentence.toLowerCase()
-            .replace(/[^a-z0-9\s]/g, ' ')
-            .split(/\s+/)
-            .filter(w => w.length > 3 && !['that', 'this', 'with', 'from', 'have', 'been', 'were', 'they', 'then', 'when', 'also', 'into', 'will', 'would', 'could', 'should'].includes(w));
-          const slug = words.slice(0, 4).join('-');
-          if (!slug) continue;
-          const topic = `auto:${slug}`;
-          const key = `cachly:lesson:best:${topic}`;
-
-          // Don't overwrite existing successful lessons
-          const existing = await redis.get(key);
-          if (existing) {
-            try {
-              const ex = JSON.parse(existing) as { outcome: string };
-              if (ex.outcome === 'success') continue;
-            } catch { /* ignore */ }
-          }
-
-          const lesson = {
-            topic,
-            outcome: 'success',
-            what_worked: sentence,
-            context: `Auto-learned from session summary. Full summary: ${summary.slice(0, 300)}`,
-            severity: 'minor',
-            ts: now.toISOString(),
-            recall_count: 0,
-            auto_learned: true,
-            session_ts: now.toISOString(),
-            version: 2,
-          };
-          await redis.set(key, JSON.stringify(lesson));
-          // 90-day TTL for auto-learned lessons
-          await redis.expire(key, 90 * 86400);
-          autoLearned.push(topic);
-        }
-
-        // Also store a lesson per changed file area if files were changed
-        if (files_changed.length > 0) {
-          const areas = [...new Set(files_changed.map(f => f.split('/').slice(0, 2).join('/')))].slice(0, 3);
-          for (const area of areas) {
-            const slug = area.replace(/[^a-z0-9]/gi, '-').toLowerCase().replace(/-+/g, '-').slice(0, 30);
-            const topic = `auto:changed:${slug}`;
-            const key = `cachly:lesson:best:${topic}`;
-            const lesson = {
-              topic,
-              outcome: 'success',
-              what_worked: `Files changed in ${area}: ${files_changed.filter(f => f.startsWith(area.split('/')[0])).slice(0, 5).join(', ')}`,
-              context: summary.slice(0, 200),
-              severity: 'minor',
-              ts: now.toISOString(),
-              recall_count: 0,
-              auto_learned: true,
-              version: 2,
-            };
-            await redis.set(key, JSON.stringify(lesson));
-            await redis.expire(key, 90 * 86400);
-            autoLearned.push(topic);
-          }
-        }
-      } catch { /* auto-learn errors must never break session_end */ }
-
-      // ── 🌿 Ambient Git Learning ────────────────────────────────────────────────
-      // Read git commits since session start → auto-learn each meaningful commit.
-      const ambientLearned: string[] = [];
-      if (workspace_path) {
-        try {
-          // Get the session start time (stored by session_start)
-          const sessionStartTs = currentRaw
-            ? (() => { try { return (JSON.parse(currentRaw) as { started?: string }).started ?? ''; } catch { return ''; } })()
-            : '';
-          const sinceArg = sessionStartTs ? `--since="${sessionStartTs}"` : '--since="1 hour ago"';
-          const gitOut = execSync(
-            `git -C "${workspace_path}" log ${sinceArg} --oneline --format="%H|||%s|||%ai"`,
-            { timeout: 5000, encoding: 'utf8', stdio: ['ignore', 'pipe', 'ignore'] },
-          ).trim();
-          if (gitOut) {
-            const commitActionRe = /\b(fix|add|remove|refactor|migrate|update|resolve|implement|improve|optimize|configure|create|delete|disable|enable|debug|patch|upgrade|build|rewrite|deploy|feat|chore|docs|test|perf|ci)\b/i;
-            for (const line of gitOut.split('\n').slice(0, 10)) {
-              const [hash, msg, dateStr] = line.split('|||');
-              if (!msg || !commitActionRe.test(msg)) continue;
-              const slug = msg
-                .toLowerCase().replace(/^(fix|feat|chore|docs|test|ci|perf|refactor|build|revert)[:(\s]/i, '')
-                .replace(/[^a-z0-9]+/g, '-').replace(/^-+|-+$/g, '').slice(0, 40);
-              if (!slug) continue;
-              const topic = `git:${slug}`;
-              const key = `cachly:lesson:best:${topic}`;
-              const existing = await redis.get(key);
-              if (existing) continue; // don't overwrite existing
-              const commitLesson = {
-                topic,
-                outcome: 'success' as const,
-                what_worked: msg.slice(0, 200),
-                context: `Auto-learned from git commit ${(hash ?? '').slice(0, 7)} at ${dateStr ?? ''} in ${workspace_path}`,
-                severity: 'minor' as const,
-                ts: now.toISOString(),
-                recall_count: 0,
-                auto_learned: true,
-                source: 'ambient-git',
-                version: 3,
-              };
-              await redis.set(key, JSON.stringify(commitLesson));
-              await redis.expire(key, 60 * 86400); // 60 day TTL for git lessons
-              ambientLearned.push(topic);
-            }
-          }
-        } catch { /* git not available or not a repo — silent skip */ }
-      }
-
-      const durationStr = durationMin !== undefined ? ` · ${durationMin} min` : '';
-      const totalAutoLearned = autoLearned.length + ambientLearned.length;
-
-      // ── Shareable Session Summary Card ────────────────────────────────────────
-      // Generated after each session so the user can share their progress.
-      const tweetLines: string[] = [];
-      if (durationMin !== undefined && durationMin > 0) tweetLines.push(`⏱ ${durationMin} min session`);
-      if (totalAutoLearned > 0) tweetLines.push(`🧠 ${totalAutoLearned} lessons saved to Brain`);
-      if (files_changed.length > 0) tweetLines.push(`📁 ${files_changed.length} file${files_changed.length > 1 ? 's' : ''} changed`);
-      const tweetBody = tweetLines.length > 0
-        ? `${tweetLines.join(' · ')}\n\nMy AI Brain remembers this so I never repeat it. @cachlydev\ncachly.dev`
-        : `Session saved to my AI Brain. No more re-explaining this tomorrow. @cachlydev\ncachly.dev`;
-      const tweetURL = `https://twitter.com/intent/tweet?text=${encodeURIComponent(tweetBody)}`;
-
-      const sessionCard = [
-        ``,
-        `┌─────────────────────────────────────────────┐`,
-        `│  🧠 Session Summary Card                    │`,
-        `│  Share your progress — cached forever       │`,
-        `├─────────────────────────────────────────────┤`,
-        durationMin !== undefined ? `│  ⏱  Duration   : ${String(durationMin + ' min').padEnd(26)}│` : '',
-        totalAutoLearned > 0     ? `│  📚 Learned    : ${String(totalAutoLearned + ' lessons').padEnd(26)}│` : '',
-        files_changed.length > 0 ? `│  📁 Changed    : ${String(files_changed.length + ' file' + (files_changed.length > 1 ? 's' : '')).padEnd(26)}│` : '',
-        `├─────────────────────────────────────────────┤`,
-        `│  📣 Share: ${tweetURL.slice(0, 34).padEnd(34)}│`,
-        `└─────────────────────────────────────────────┘`,
-      ].filter(l => l !== '').join('\n');
-
-      return [
-        `✅ **Session saved**${durationStr}`,
-        ``,
-        `📋 **Summary:** ${summary}`,
-        files_changed.length > 0 ? `📁 **Files changed:** ${files_changed.map(f => `\`${f}\``).join(', ')}` : '',
-        lessons_learned !== undefined ? `🧠 **Lessons stored:** ${lessons_learned}` : '',
-        autoLearned.length > 0 ? `🤖 **Auto-learned:** ${autoLearned.length} lessons extracted from summary (${autoLearned.slice(0, 3).map(t => `\`${t}\``).join(', ')}${autoLearned.length > 3 ? '…' : ''})` : '',
-        ambientLearned.length > 0 ? `🌿 **Ambient git learning:** ${ambientLearned.length} commit${ambientLearned.length > 1 ? 's' : ''} auto-learned (${ambientLearned.slice(0, 3).map(t => `\`${t}\``).join(', ')}${ambientLearned.length > 3 ? '…' : ''})` : '',
-        sessionCard,
-        ``,
-        `💡 Next session: \`session_start(focus="...")\` to see this summary.`,
-      ].filter(l => l !== '').join('\n');
-    }
-
-    // ── session_ping — lightweight checkpoint ─────────────────────────────────
-    case 'session_ping': {
-      const {
-        instance_id,
-        task,
-        files_touched = [],
-        next_step = '',
-        provider = '',
-      } = args as {
-        instance_id: string;
-        task: string;
-        files_touched?: string[];
-        next_step?: string;
-        provider?: string;
-      };
-
-      const redis = await getConnection(instance_id);
-      const checkpoint = {
-        ts: new Date().toISOString(),
-        task,
-        files_touched,
-        next_step,
-        provider,
-      };
-
-      // Store as the latest checkpoint — session_start reads this when no session_end found
-      await redis.set('cachly:session:checkpoint', JSON.stringify(checkpoint), 'EX', 86400 * 3); // 3-day TTL
-
-      // Also keep a short rolling log (last 20 checkpoints for history)
-      await redis.lpush('cachly:session:checkpoint:log', JSON.stringify(checkpoint));
-      await redis.ltrim('cachly:session:checkpoint:log', 0, 19);
-
-      const providerStr = provider ? ` [${provider}]` : '';
-      const filesStr = files_touched.length > 0 ? ` · ${files_touched.length} file${files_touched.length > 1 ? 's' : ''} touched` : '';
-      const nextStr = next_step ? `\n📍 **Next step:** ${next_step}` : '';
-
-      return [
-        `📌 **Checkpoint saved**${providerStr} — ${new Date().toLocaleTimeString()}`,
-        `🔨 **Working on:** ${task}${filesStr}`,
-        nextStr,
-        ``,
-        `💡 If you switch providers, \`session_start\` will show this checkpoint automatically.`,
-      ].filter(l => l !== '').join('\n');
-    }
-
-    // ── session_handoff — cross-window continuity ─────────────────────────────
-    case 'session_handoff': {
-      const {
-        instance_id,
-        completed_tasks = [],
-        remaining_tasks = [],
-        files_changed = [],
-        instructions = '',
-        context_summary = '',
-        blocked_on = '',
-      } = args as {
-        instance_id: string;
-        completed_tasks: string[];
-        remaining_tasks: string[];
-        files_changed?: { path: string; status: string; description?: string }[];
-        instructions?: string;
-        context_summary?: string;
-        blocked_on?: string;
-      };
-
-      const redis = await getConnection(instance_id);
-      const now = new Date();
-
-      const handoff = {
-        ts: now.toISOString(),
-        completed_tasks,
-        remaining_tasks,
-        files_changed,
-        instructions,
-        context_summary,
-        blocked_on,
-      };
-
-      // Store handoff — never expires until next handoff overwrites it
-      await redis.set('cachly:session:handoff', JSON.stringify(handoff));
-
-      // Also append to history
-      await redis.lpush('cachly:session:handoff:history', JSON.stringify(handoff));
-      await redis.ltrim('cachly:session:handoff:history', 0, 19);
-
-      const totalTasks = completed_tasks.length + remaining_tasks.length;
-      const pct = totalTasks > 0 ? Math.round((completed_tasks.length / totalTasks) * 100) : 0;
-      const brokenFiles = files_changed.filter(f => f.status === 'broken' || f.status === 'partial');
-
-      return [
-        `🤝 **Handoff saved** — ${completed_tasks.length}/${totalTasks} tasks done (${pct}%)`,
-        ``,
-        completed_tasks.length > 0 ? `✅ **Completed:**\n${completed_tasks.map(t => `  - ${t}`).join('\n')}` : '',
-        remaining_tasks.length > 0 ? `\n⏳ **Remaining for next window:**\n${remaining_tasks.map(t => `  - ${t}`).join('\n')}` : '',
-        brokenFiles.length > 0 ? `\n⚠️ **Needs attention:** ${brokenFiles.map(f => `\`${f.path}\` (${f.status})`).join(', ')}` : '',
-        blocked_on ? `\n🚫 **Blocked on:** ${blocked_on}` : '',
-        instructions ? `\n📝 **Instructions:** ${instructions}` : '',
-        ``,
-        `💡 The next \`session_start\` will include this handoff automatically.`,
-      ].filter(l => l !== '').join('\n');
-    }
-
-    // ── auto_learn_session ────────────────────────────────────────────────────
-    case 'auto_learn_session': {
-      const { instance_id, observations } = args as {
-        instance_id: string;
-        observations: { action: string; outcome: string; details?: string; topic?: string; severity?: string }[];
-      };
-      const redis = await getConnection(instance_id);
-      const stored: string[] = [];
-      const skipped: string[] = [];
-
-      for (const obs of observations) {
-        // Auto-generate topic from action if not provided
-        const rawTopic = obs.topic ?? obs.action
-          .toLowerCase()
-          .replace(/[^a-z0-9:\-_\s]/g, '')
-          .trim()
-          .split(/\s+/)
-          .slice(0, 4)
-          .join('-');
-        const topic = rawTopic.includes(':') ? rawTopic : `auto:${rawTopic}`;
-        const key = `cachly:lesson:best:${topic}`;
-
-        // Only overwrite if this is a success and existing is failure, or topic is new
-        const existing = await redis.get(key);
-        if (existing) {
-          const existingLesson = JSON.parse(existing) as { outcome: string };
-          if (existingLesson.outcome === 'success' && obs.outcome !== 'success') {
-            skipped.push(topic);
-            continue;
-          }
-        }
-
-        const lesson = {
-          topic,
-          outcome: obs.outcome,
-          what_worked: obs.outcome === 'success' ? obs.action : (obs.details ?? obs.action),
-          what_failed: obs.outcome === 'failure' ? obs.action : undefined,
-          context: obs.details,
-          severity: obs.severity ?? 'minor',
-          ts: new Date().toISOString(),
-          recall_count: 0,
-          auto_learned: true,
-          version: 2,
-        };
-
-        await redis.set(key, JSON.stringify(lesson));
-        stored.push(`${obs.outcome === 'success' ? '✅' : obs.outcome === 'partial' ? '⚠️' : '❌'} \`${topic}\``);
-      }
-
-      const lines = [
-        `🤖 **Auto-learn complete**: ${stored.length} stored, ${skipped.length} skipped`,
-        '',
-      ];
-      if (stored.length > 0) lines.push('**Stored:**', ...stored.map(s => '  ' + s), '');
-      if (skipped.length > 0) lines.push(`**Skipped** (better lesson already exists): ${skipped.map(t => `\`${t}\``).join(', ')}`);
-      return lines.join('\n');
-    }
-
-    // ── brain_who_knows ───────────────────────────────────────────────────────
-    // Phase 3A: Org-wide people intelligence. Returns ranked list of contributors
-    // whose authored edges in the CKG match the query topic.
-    case 'brain_who_knows': {
-      const {
-        instance_id,
-        topic,
-        limit = 10,
-      } = args as { instance_id: string; topic: string; limit?: number };
-
-      if (!topic || typeof topic !== 'string' || !topic.trim()) {
-        return '⚠️ `brain_who_knows` requires a non-empty `topic`. Example: `brain_who_knows(topic="kubernetes deployment")`.';
-      }
-      const safeLimit = Math.max(1, Math.min(Number(limit) || 10, 50));
-
-      const redis = await getConnection(instance_id);
-
-      // Step 1: Find candidate concept node IDs matching the query
-      const querySlug = ckgSlug(topic);
-      const qTokens = tokenize(topic).filter((t: string) => t.length >= 3).slice(0, 5);
-
-      const candidateConceptIds = new Set<string>([querySlug]);
-      for (const token of qTokens.slice(0, 3)) {
-        const nodeStream = redis.scanStream({ match: `cachly:ckg:node:*${token}*`, count: 30 });
-        await new Promise<void>((res, rej) => {
-          nodeStream.on('data', (batch: string[]) => {
-            for (const k of batch.slice(0, 10)) {
-              const nodeId = k.replace('cachly:ckg:node:', '');
-              if (!nodeId.startsWith('person:') && !nodeId.startsWith('file:'))
-                candidateConceptIds.add(nodeId);
-            }
-          });
-          nodeStream.on('end', res);
-          nodeStream.on('error', rej);
-        });
-      }
-
-      // Step 2: Traverse authored edges (person → concept) to find experts
-      type PersonScore = {
-        handle: string; lessonCount: number; totalConfidence: number;
-        domains: Set<string>; lastActive: string;
-      };
-      const personScores = new Map<string, PersonScore>();
-
-      for (const conceptId of [...candidateConceptIds].slice(0, 25)) {
-        const inboundEdgeKeys = await redis.smembers(`cachly:ckg:idx:to:${conceptId}`);
-        for (const ek of inboundEdgeKeys) {
-          const er = await redis.get(ek);
-          if (!er) continue;
-          const edge = safeJsonParse<CKGEdge | null>(er, null);
-          if (!edge || edge.edgeType !== 'authored') continue;
-
-          const personNodeRaw = await redis.get(`cachly:ckg:node:${edge.from}`);
-          if (!personNodeRaw) continue;
-          const personNode = safeJsonParse<PersonNode | null>(personNodeRaw, null);
-          if (!personNode || personNode.type !== 'person') continue;
-
-          const existing = personScores.get(edge.from);
-          if (existing) {
-            existing.lessonCount++;
-            existing.totalConfidence += edge.confidence;
-            existing.domains.add(personNode.domain);
-            if (edge.last_updated > existing.lastActive) existing.lastActive = edge.last_updated;
-          } else {
-            personScores.set(edge.from, {
-              handle: personNode.handle,
-              lessonCount: 1,
-              totalConfidence: edge.confidence,
-              domains: new Set([personNode.domain]),
-              lastActive: personNode.last_active ?? edge.last_updated,
-            });
-          }
-        }
-      }
-
-      if (personScores.size === 0) {
-        return [
-          `## 👥 Who Knows About \`${topic}\`?`,
-          ``,
-          `No attributed lessons found yet.`,
-          ``,
-          `Knowledge attribution builds automatically — add \`author="name"\` to \`learn_from_attempts\` calls:`,
-          `\`learn_from_attempts(topic="${topic}", author="your-handle", ...)\``,
-        ].join('\n');
-      }
-
-      // Step 3: Sort by expertise score (lesson count × avg confidence)
-      const ranked = [...personScores.entries()]
-        .map(([, s]) => ({
-          handle: s.handle,
-          lessonCount: s.lessonCount,
-          avgConfidence: s.totalConfidence / s.lessonCount,
-          domains: [...s.domains].slice(0, 3).join(', '),
-          lastActive: s.lastActive,
-        }))
-        .sort((a, b) => (b.lessonCount * b.avgConfidence) - (a.lessonCount * a.avgConfidence))
-        .slice(0, safeLimit);
-
-      const lines = [
-        `## 👥 Who Knows About \`${topic}\`?`,
-        ``,
-        `**${ranked.length}** contributor${ranked.length !== 1 ? 's' : ''} with relevant lessons:`,
-        ``,
-      ];
-
-      for (const [i, p] of ranked.entries()) {
-        const medal = i === 0 ? '🥇' : i === 1 ? '🥈' : i === 2 ? '🥉' : `${i + 1}.`;
-        const conf = (p.avgConfidence * 100).toFixed(0);
-        const daysSince = Math.floor((Date.now() - new Date(p.lastActive).getTime()) / 86400000);
-        const recency = daysSince < 1 ? 'today' : daysSince < 7 ? `${daysSince}d ago` : daysSince < 30 ? `${Math.floor(daysSince / 7)}w ago` : `${Math.floor(daysSince / 30)}mo ago`;
-        lines.push(
-          `${medal} **${p.handle}** — ${p.lessonCount} lesson${p.lessonCount !== 1 ? 's' : ''} · ${conf}% confidence · last active ${recency}`,
-          `   _domains: ${p.domains}_`,
-          ``,
-        );
-      }
-
-      lines.push(
-        `---`,
-        `_Attribution grows with every \`learn_from_attempts(author="...", ...)\` call._`,
-      );
-
-      return lines.join('\n');
-    }
-
-    // ── brain_file_map ────────────────────────────────────────────────────────
-    // Phase 3B: "What do we know about these files?" — experts + lessons per file.
-    case 'brain_file_map': {
-      const {
-        instance_id,
-        file_paths = [],
-      } = args as { instance_id: string; file_paths: string[] };
-
-      // Defensive: accept only a non-empty array of non-empty strings.
-      const cleanPaths = (Array.isArray(file_paths) ? file_paths : [])
-        .filter((p): p is string => typeof p === 'string' && p.trim().length > 0)
-        .map(p => p.trim());
-      if (!cleanPaths.length) return '⚠️ Pass at least one file path. Example: `brain_file_map(file_paths=["src/auth/jwt.ts"])`.';
-
-      const redis = await getConnection(instance_id);
-      const lines: string[] = [
-        `## 📁 File Knowledge Map`,
-        ``,
-        `_What cachly knows about the files you're about to touch:_`,
-        ``,
-      ];
-
-      for (const fp of cleanPaths.slice(0, 10)) {
-        lines.push(`### \`${fp}\``);
-
-        // ── Experts via CKG file nodes ─────────────────────────────────────
-        const fileId = `file:${ckgSlug(fp)}`;
-        const inbound = await redis.smembers(`cachly:ckg:idx:to:${fileId}`);
-        type ExpertEntry = { handle: string; touches: number; lastSeen: string };
-        const experts: ExpertEntry[] = [];
-        for (const ek of inbound) {
-          const er = await redis.get(ek);
-          if (!er) continue;
-          const edge = safeJsonParse<CKGEdge | null>(er, null);
-          if (!edge || edge.edgeType !== 'touched') continue;
-          const personRaw = await redis.get(`cachly:ckg:node:${edge.from}`);
-          if (!personRaw) continue;
-          const pn = safeJsonParse<PersonNode | null>(personRaw, null);
-          if (!pn || pn.type !== 'person') continue;
-          experts.push({ handle: pn.handle, touches: edge.trials, lastSeen: edge.last_updated });
-        }
-        experts.sort((a, b) => b.touches - a.touches);
-
-        if (experts.length > 0) {
-          const expLine = experts.slice(0, 5).map((e, i) => {
-            const medal = i === 0 ? '🥇' : i === 1 ? '🥈' : i === 2 ? '🥉' : `${i + 1}.`;
-            const days = Math.floor((Date.now() - new Date(e.lastSeen).getTime()) / 86400000);
-            const ago = days < 1 ? 'today' : days < 7 ? `${days}d ago` : `${Math.floor(days / 7)}w ago`;
-            return `${medal} **${e.handle}** (${e.touches}× · ${ago})`;
-          }).join(' · ');
-          lines.push(`**Experts:** ${expLine}`);
-        } else {
-          lines.push(`**Experts:** _None yet — add \`author\` to \`learn_from_attempts\` calls_`);
-        }
-
-        // ── Related lessons via BM25 on filename tokens ────────────────────
-        const tokens = fp.replace(/[^a-z0-9]/gi, ' ').split(/\s+/).filter(t => t.length >= 4);
-        const lessonQuery = tokens.slice(0, 3).join(' ');
-        const relatedLessons: string[] = [];
-        if (lessonQuery) {
-          try {
-            const hits = await keywordSearch(redis, ['cachly:lesson:best:*'], lessonQuery, 4);
-            for (const h of hits) {
-              const ld = safeJsonParse<{
-                topic?: string; outcome?: string; what_worked?: string;
-                severity?: string; author?: string; visibility?: string;
-              }>(h.content, {});
-              if (ld.visibility === 'private') continue;
-              const emoji = ld.outcome === 'success' ? '✅' : ld.outcome === 'partial' ? '⚠️' : '❌';
-              const sev = ld.severity === 'critical' ? ' 🔴' : ld.severity === 'major' ? ' 🟡' : '';
-              const by = ld.author ? ` · 👤 ${ld.author}` : '';
-              const topic = h.key.replace('cachly:lesson:best:', '');
-              relatedLessons.push(`  - ${emoji} \`${topic}\`${sev}${by} — ${(ld.what_worked ?? '').slice(0, 100)}`);
-            }
-          } catch { /* non-critical */ }
-        }
-        // Also: exact file_paths match via lesson scan
-        if (relatedLessons.length < 2) {
-          const allLessonKeys = await scanKeys(redis, 'cachly:lesson:best:*', { max: 1500 });
-          for (const k of allLessonKeys.slice(0, 300)) {
-            const raw = await redis.get(k);
-            if (!raw) continue;
-            const ld = safeJsonParse<{
-              topic?: string; outcome?: string; what_worked?: string;
-              severity?: string; author?: string; visibility?: string; file_paths?: string[];
-            }>(raw, {});
-            if (ld.visibility === 'private') continue;
-            if (!(ld.file_paths ?? []).includes(fp)) continue;
-            const topic = k.replace('cachly:lesson:best:', '');
-            if (relatedLessons.some(l => l.includes(`\`${topic}\``))) continue;
-            const emoji = ld.outcome === 'success' ? '✅' : ld.outcome === 'partial' ? '⚠️' : '❌';
-            const sev = ld.severity === 'critical' ? ' 🔴' : ld.severity === 'major' ? ' 🟡' : '';
-            const by = ld.author ? ` · 👤 ${ld.author}` : '';
-            relatedLessons.push(`  - ${emoji} \`${topic}\`${sev}${by} — ${(ld.what_worked ?? '').slice(0, 100)}`);
-          }
-        }
-
-        if (relatedLessons.length > 0) {
-          lines.push(`**Related lessons:**`);
-          lines.push(...relatedLessons.slice(0, 5));
-        } else {
-          lines.push(`**Related lessons:** _None yet_`);
-        }
-        lines.push(``);
-      }
-
-      lines.push(
-        `---`,
-        `_Attribution builds with \`learn_from_attempts(file_paths=[...], author="...")\`. Run this before committing to track what changed and why._`,
-      );
-      return lines.join('\n');
-    }
-
-    // ── team_expertise_map ────────────────────────────────────────────────────
-    // Phase 3B: Full team expertise overview — who knows what, at a glance.
-    case 'team_expertise_map': {
-      const {
-        instance_id,
-        top_n = 20,
-      } = args as { instance_id: string; top_n?: number };
-
-      const redis = await getConnection(instance_id);
-
-      // Scan all person nodes (capped + timed out so a huge keyspace can't hang the agent)
-      const personKeys = await scanKeys(redis, 'cachly:ckg:node:person:*', { max: 2000 });
-
-      if (personKeys.length === 0) {
-        return [
-          `## 🗺️ Team Expertise Map`,
-          ``,
-          `No contributors yet.`,
-          ``,
-          `Add \`author="handle"\` to any \`learn_from_attempts\` call to start building the map.`,
-        ].join('\n');
-      }
-
-      type PersonEntry = { handle: string; lessonCount: number; domains: Map<string, number>; lastActive: string };
-      const people: PersonEntry[] = [];
-
-      for (const pk of personKeys) {
-        const raw = await redis.get(pk);
-        if (!raw) continue;
-        const pn = safeJsonParse<PersonNode | null>(raw, null);
-        if (!pn || pn.type !== 'person') continue;
-
-        // Count authored lessons + collect domains from authored edges
-        const outEdgeKeys = await redis.smembers(`cachly:ckg:idx:from:${pn.id}`);
-        const domainCounts = new Map<string, number>();
-        let lessonCount = 0;
-        for (const ek of outEdgeKeys) {
-          const er = await redis.get(ek);
-          if (!er) continue;
-          const edge = safeJsonParse<CKGEdge | null>(er, null);
-          if (!edge || edge.edgeType !== 'authored') continue;
-          lessonCount++;
-          const domain = edge.to.split(':')[0] ?? 'unknown';
-          domainCounts.set(domain, (domainCounts.get(domain) ?? 0) + 1);
-        }
-        people.push({ handle: pn.handle, lessonCount, domains: domainCounts, lastActive: pn.last_active });
-      }
-
-      people.sort((a, b) => b.lessonCount - a.lessonCount);
-      const topPeople = people.slice(0, top_n);
-
-      const lines = [
-        `## 🗺️ Team Expertise Map`,
-        ``,
-        `**${topPeople.length}** contributor${topPeople.length !== 1 ? 's' : ''} tracked:`,
-        ``,
-        `| # | Handle | Lessons | Top domains | Last active |`,
-        `|---|--------|---------|-------------|-------------|`,
-      ];
-
-      for (const [i, p] of topPeople.entries()) {
-        const medal = i === 0 ? '🥇' : i === 1 ? '🥈' : i === 2 ? '🥉' : `${i + 1}`;
-        const days = Math.floor((Date.now() - new Date(p.lastActive).getTime()) / 86400000);
-        const ago = days < 1 ? 'today' : days < 7 ? `${days}d ago` : days < 30 ? `${Math.floor(days / 7)}w ago` : `${Math.floor(days / 30)}mo ago`;
-        const domainStr = [...p.domains.entries()].sort((a, b) => b[1] - a[1]).slice(0, 3).map(([d, c]) => `${d}(${c})`).join(', ') || '—';
-        lines.push(`| ${medal} | **${p.handle}** | ${p.lessonCount} | ${domainStr} | ${ago} |`);
-      }
-
-      lines.push(
-        ``,
-        `---`,
-        `_Use \`brain_who_knows(topic="...")\` to find the expert for a specific problem._`,
-      );
-      return lines.join('\n');
-    }
-
-    // ── skill_gaps ────────────────────────────────────────────────────────────
-    // Phase 3C: Show knowledge blind spots — domains/files with no success lessons,
-    // no attribution, or only failures.
-    case 'skill_gaps': {
-      const {
-        instance_id,
-        min_failures = 1,
-      } = args as { instance_id: string; min_failures?: number };
-
-      const redis = await getConnection(instance_id);
-
-      // Scan all best lessons (capped + timed out)
-      const lessonKeys = await scanKeys(redis, 'cachly:lesson:best:*', { max: 3000 });
-
-      type DomainStats = { successes: number; failures: number; attributed: number; topics: string[] };
-      const domainMap = new Map<string, DomainStats>();
-      let totalAttributed = 0;
-      let totalPrivate = 0;
-
-      for (const k of lessonKeys) {
-        const raw = await redis.get(k);
-        if (!raw) continue;
-        const l = safeJsonParse<{
-          topic?: string; outcome?: string; author?: string; visibility?: string;
-        }>(raw, {});
-        const domain = (l.topic ?? '').split(':')[0] ?? 'unknown';
-        const s = domainMap.get(domain) ?? { successes: 0, failures: 0, attributed: 0, topics: [] };
-        if (l.outcome === 'success' || l.outcome === 'partial') s.successes++;
-        else s.failures++;
-        if (l.author) { s.attributed++; totalAttributed++; }
-        if (l.visibility === 'private') totalPrivate++;
-        s.topics.push(l.topic ?? '');
-        domainMap.set(domain, s);
-      }
-
-      // Identify gaps
-      type Gap = { domain: string; type: string; severity: string; detail: string };
-      const gaps: Gap[] = [];
-
-      for (const [domain, s] of domainMap) {
-        if (s.failures >= min_failures && s.successes === 0) {
-          gaps.push({
-            domain, type: 'unresolved failures',
-            severity: s.failures >= 3 ? '🔴 critical' : '🟡 warn',
-            detail: `${s.failures} failure${s.failures !== 1 ? 's' : ''}, 0 solutions — use \`learn_from_attempts\` to store what fixed it`,
-          });
-        }
-        if (s.successes + s.failures >= 3 && s.attributed === 0) {
-          gaps.push({
-            domain, type: 'no attribution',
-            severity: '🔵 info',
-            detail: `${s.successes + s.failures} lessons with no \`author\` field — attribution lost, \`brain_who_knows\` can't help here`,
-          });
-        }
-      }
-
-      const lines = [
-        `## 🔍 Skill Gaps — Knowledge Blind Spots`,
-        ``,
-        `**${lessonKeys.length}** lessons scanned · **${totalAttributed}** attributed · **${gaps.length}** gap${gaps.length !== 1 ? 's' : ''} found`,
-        totalPrivate > 0 ? `_(${totalPrivate} private lessons excluded from gap analysis)_` : '',
-        ``,
-      ].filter(Boolean);
-
-      if (gaps.length === 0) {
-        lines.push(`✅ No significant knowledge gaps detected. All domains with failures have at least one success lesson.`);
-      } else {
-        const bySev = [
-          ...gaps.filter(g => g.severity.includes('critical')),
-          ...gaps.filter(g => g.severity.includes('warn')),
-          ...gaps.filter(g => g.severity.includes('info')),
-        ];
-        for (const g of bySev.slice(0, 15)) {
-          lines.push(`${g.severity} **\`${g.domain}\`** — ${g.type}`);
-          lines.push(`  ${g.detail}`);
-          lines.push(``);
-        }
-        if (gaps.length > 15) lines.push(`_…and ${gaps.length - 15} more gaps. Fix critical ones first._`);
-      }
-
-      lines.push(
-        `---`,
-        `_Fix gaps with \`learn_from_attempts(topic="${gaps[0]?.domain ?? 'your:topic'}", outcome="success", what_worked="...")\`_`,
-        `_Add \`author="name"\` to every call to enable \`brain_who_knows\` + \`team_expertise_map\`._`,
-      );
-
-      return lines.join('\n');
-    }
-
-    // ── brain_coverage ────────────────────────────────────────────────────────
-    // Phase 3C: Knowledge-coverage health score for the repository.
-    case 'brain_coverage': {
-      const {
-        instance_id,
-        repo_path = '.',
-      } = args as { instance_id: string; repo_path?: string };
-
-      const redis = await getConnection(instance_id);
-
-      // Count lessons, person + file nodes (each capped + timed out, gathered in parallel)
-      const [lessonKeys, personKeys, fileKeys] = await Promise.all([
-        scanKeys(redis, 'cachly:lesson:best:*', { max: 5000 }),
-        scanKeys(redis, 'cachly:ckg:node:person:*', { max: 2000 }),
-        scanKeys(redis, 'cachly:ckg:node:file:*', { max: 5000 }),
-      ]);
-
-      // Get git file count for comparison (optional — non-critical if not a git repo)
-      let gitFileCount = 0;
-      let _gitFileCountLabel = 'n/a';
-      try {
-        const { execSync } = await import('node:child_process');
-        const { resolve } = await import('node:path');
-        const repoDir = resolve(repo_path);
-        const output = execSync('git ls-files', { cwd: repoDir, encoding: 'utf-8', stdio: 'pipe' });
-        gitFileCount = output.trim().split('\n').filter(Boolean).length;
-        _gitFileCountLabel = String(gitFileCount);
-      } catch { /* not a git repo — skip */ }
-
-      // Lesson quality breakdown
-      let successCount = 0; let failureCount = 0; let attributedCount = 0;
-      const domainSet = new Set<string>();
-      for (const k of lessonKeys) {
-        const raw = await redis.get(k);
-        if (!raw) continue;
-        const l = safeJsonParse<{ outcome?: string; author?: string; topic?: string }>(raw, {});
-        if (l.outcome === 'success' || l.outcome === 'partial') successCount++;
-        else failureCount++;
-        if (l.author) attributedCount++;
-        domainSet.add((l.topic ?? '').split(':')[0] ?? 'unknown');
-      }
-
-      const fileCoverage = gitFileCount > 0
-        ? `${Math.round((fileKeys.length / gitFileCount) * 100)}% (${fileKeys.length}/${gitFileCount} files)`
-        : `${fileKeys.length} files tracked`;
-
-      const attributionPct = lessonKeys.length > 0
-        ? Math.round((attributedCount / lessonKeys.length) * 100)
-        : 0;
-
-      const overallScore = Math.round(
-        (Math.min(1, lessonKeys.length / 50) * 25) +          // lesson volume (25 pts)
-        (Math.min(1, successCount / Math.max(1, lessonKeys.length)) * 25) + // success ratio (25 pts)
-        (Math.min(1, attributionPct / 100) * 25) +            // attribution completeness (25 pts)
-        (Math.min(1, personKeys.length / 5) * 25)             // team engagement (25 pts)
-      );
-
-      const scoreEmoji = overallScore >= 80 ? '🟢' : overallScore >= 50 ? '🟡' : '🔴';
-
-      return [
-        `## 📊 Brain Coverage Report`,
-        ``,
-        `${scoreEmoji} **Overall score: ${overallScore}/100**`,
-        ``,
-        `| Metric | Value |`,
-        `|--------|-------|`,
-        `| Total lessons | **${lessonKeys.length}** (${successCount} success · ${failureCount} failure) |`,
-        `| Problem domains | **${domainSet.size}** |`,
-        `| Attribution | **${attributedCount}/${lessonKeys.length}** (${attributionPct}%) |`,
-        `| Contributors | **${personKeys.length}** tracked in knowledge graph |`,
-        `| File coverage | **${fileCoverage}** |`,
-        ``,
-        `**Score breakdown (each 0–25):**`,
-        `- Lesson volume: **${Math.round(Math.min(1, lessonKeys.length / 50) * 25)}/25** _(target: 50+ lessons)_`,
-        `- Success ratio: **${Math.round(Math.min(1, successCount / Math.max(1, lessonKeys.length)) * 25)}/25** _(target: all lessons resolved)_`,
-        `- Attribution: **${Math.round(Math.min(1, attributionPct / 100) * 25)}/25** _(target: 100% attributed)_`,
-        `- Team engagement: **${Math.round(Math.min(1, personKeys.length / 5) * 25)}/25** _(target: 5+ contributors)_`,
-        ``,
-        `---`,
-        `_Boost score: run \`brain_from_git\` · add \`author\` to \`learn_from_attempts\` · fix gaps with \`skill_gaps\`_`,
-      ].join('\n');
-    }
-
-    // ── brain_metrics ─────────────────────────────────────────────────────────
-    // The three decisive metrics (PROGRESS.md §3 / VISION_10X.md §4):
-    //   1. Time-to-first-recall  — onboarding friction
-    //   2. Recall-lift           — the moat proof (vs. raw BM25)
-    //   3. Team-knowledge-reuse  — the value only a shared brain delivers
-    case 'brain_metrics': {
-      const { instance_id } = args as { instance_id: string };
-      const redis = await getConnection(instance_id);
-
-      const [bornAt, firstRecallAt, recallsTotalRaw, crossAuthorRaw, timeSavedRaw, reusePairs] =
-        await Promise.all([
-          redis.get(`cachly:stats:born_at:${instance_id}`).catch(() => null),
-          redis.get(`cachly:stats:first_recall_at:${instance_id}`).catch(() => null),
-          redis.get(`cachly:stats:recalls_total:${instance_id}`).catch(() => null),
-          redis.get(`cachly:stats:cross_author_recalls:${instance_id}`).catch(() => null),
-          redis.get(`cachly:stats:time_saved_mins:${instance_id}`).catch(() => null),
-          redis.smembers(`cachly:stats:reuse_pairs:${instance_id}`).catch(() => [] as string[]),
-        ]);
-
-      // ── Metric 1: Time-to-first-recall ──────────────────────────────────────
-      let ttfrLine: string;
-      if (bornAt && firstRecallAt) {
-        const ms = new Date(firstRecallAt).getTime() - new Date(bornAt).getTime();
-        const mins = Math.max(0, Math.round(ms / 60000));
-        const human = mins < 1 ? '<1 min' : mins < 60 ? `${mins} min` : `${(mins / 60).toFixed(1)} h`;
-        const target = ms <= 2 * 60000 ? '🟢 under 2 min target' : ms <= 60 * 60000 ? '🟡 over 2 min' : '🔴 over 1 h';
-        ttfrLine = `**${human}** from first lesson → first proven recall · ${target}`;
-      } else if (bornAt && !firstRecallAt) {
-        ttfrLine = `🟡 Brain has knowledge but no proven recall yet — try \`smart_recall\``;
-      } else {
-        ttfrLine = `⚪ Not enough data yet — store a lesson, then recall it`;
-      }
-
-      // ── Metric 2: Recall-lift (the moat proof) ──────────────────────────────
-      // Published headline from Cachly-Bench (CI-defended in rerank.test.ts).
-      const recallLiftLine = `**+22.2 % Precision@1**, **+10.9 % MRR**, **+8.1 % nDCG@5** vs. raw BM25`;
-
-      // ── Metric 3: Team-knowledge-reuse ──────────────────────────────────────
-      const recallsTotal = Number(recallsTotalRaw ?? 0);
-      const crossAuthor = Number(crossAuthorRaw ?? 0);
-      const reusePct = recallsTotal > 0 ? Math.round((crossAuthor / recallsTotal) * 100) : 0;
-      const reuseTarget = reusePct >= 30 ? '🟢 above 30 % target' : recallsTotal === 0 ? '⚪ no recalls yet' : '🟡 below 30 % target';
-      const timeSaved = Math.round(Number(timeSavedRaw ?? 0));
-      const timeSavedHuman = timeSaved < 60 ? `${timeSaved} min` : `${(timeSaved / 60).toFixed(1)} h`;
-
-      const lines = [
-        `## 📈 Brain Metrics — the three that decide everything`,
-        ``,
-        `### 1. Time-to-first-recall _(onboarding friction)_`,
-        ttfrLine,
-        ``,
-        `### 2. Recall-lift _(the moat proof)_`,
-        recallLiftLine,
-        `_Reproduce: \`npm run bench\` · CI-defended in \`rerank.test.ts\` · details in BENCH.md_`,
-        ``,
-        `### 3. Team-knowledge-reuse _(the value only a shared brain delivers)_`,
-        `**${reusePct}%** of proven recalls used a teammate's lesson (${crossAuthor}/${recallsTotal}) · ${reuseTarget}`,
-        reusePairs.length > 0 ? `**${reusePairs.length}** distinct reuse relationship${reusePairs.length !== 1 ? 's' : ''} across the team` : '',
-        ``,
-        `---`,
-        `⏱️ **${timeSavedHuman}** total saved not re-researching known fixes.`,
-        recallsTotal === 0
-          ? `_Tip: pass \`author="your-handle"\` to \`smart_recall\` so cross-author reuse can be tracked._`
-          : `_These numbers compound: every learned lesson and every teammate raises all three._`,
-      ].filter(Boolean);
-
-      return lines.join('\n');
-    }
-
-    // ── sync_file_changes ─────────────────────────────────────────────────────
-
-    default:
-      return null;
-  }
-}