feat(jme): Phase 4 — auto-prune low-confidence facts when vector ceiling is active

src/memory/jme.test.ts

@@ -592,6 +592,74 @@ describe("JME — consolidateAll (nightly batch)", () => {
     expect(result.factsExtracted).toBe(0);
     expect(jmeStats().turnsTotal).toBe(1); // NOT deleted — unconsumed data
   });
+
+  // ── Phase 4: auto-prune when ceiling is active ───────────────────────────
+
+  it("Phase 4: auto-prunes low-confidence facts when embedding count >= VECTOR_CEILING_WARN", async () => {
+    const { consolidateAll, VECTOR_CEILING_WARN, PHASE4_CONFIDENCE_FLOOR } =
+      await getJme();
+
+    // Insert VECTOR_CEILING_WARN low-confidence facts (confidence below Phase 4
+    // floor, ts > 30 days ago so they qualify for deletion)
+    const oldTs = Date.now() - 31 * 24 * 60 * 60 * 1000;
+    const stmt = mockDb.prepare(
+      `INSERT INTO jme_facts (source_task, ts, fact_text, category, confidence, embedding)
+       VALUES (?, ?, ?, ?, ?, ?)`,
+    );
+    // Use a tiny 4-byte blob so embedding IS NOT NULL (counted by the ceiling query)
+    const tinyEmbedding = Buffer.alloc(4);
+    for (let i = 0; i < VECTOR_CEILING_WARN; i++) {
+      stmt.run("consolidator-nightly", oldTs, `low-conf fact ${i}`, "event", 0.3, tinyEmbedding);
+      mockDb
+        .prepare(`INSERT INTO jme_facts_fts(rowid, fact_text) VALUES (last_insert_rowid(), ?)`)
+        .run(`low-conf fact ${i}`);
+    }
+
+    // Also insert one high-confidence fact that should survive
+    mockDb
+      .prepare(
+        `INSERT INTO jme_facts (source_task, ts, fact_text, category, confidence)
+         VALUES ('t-keep', ?, 'important fact', 'preference', 0.9)`,
+      )
+      .run(oldTs);
+
+    // No turns to process — consolidateAll still performs the ceiling check
+    inferMock.mockResolvedValueOnce({ content: "[]" });
+
+    const result = await consolidateAll();
+
+    // Phase 4 pruned the low-confidence facts
+    expect(result.factsPruned).toBeGreaterThanOrEqual(VECTOR_CEILING_WARN);
+    // High-confidence fact is untouched
+    const remaining = mockDb
+      .prepare(`SELECT fact_text FROM jme_facts WHERE fact_text = 'important fact'`)
+      .get();
+    expect(remaining).toBeTruthy();
+  });
+
+  it("Phase 4: factsPruned is 0 when embedding count is below VECTOR_CEILING_WARN", async () => {
+    const { consolidateAll, VECTOR_CEILING_WARN } = await getJme();
+
+    // Insert fewer facts than the ceiling threshold
+    const oldTs = Date.now() - 31 * 24 * 60 * 60 * 1000;
+    const tinyEmbedding = Buffer.alloc(4);
+    for (let i = 0; i < VECTOR_CEILING_WARN - 1; i++) {
+      mockDb
+        .prepare(
+          `INSERT INTO jme_facts (source_task, ts, fact_text, category, confidence, embedding)
+           VALUES (?, ?, ?, ?, ?, ?)`,
+        )
+        .run("t1", oldTs, `fact ${i}`, "event", 0.3, tinyEmbedding);
+    }
+
+    insertSettledTurn("task-x", "user", "hola");
+    inferMock.mockResolvedValueOnce({ content: "[]" });
+
+    const result = await consolidateAll();
+
+    // Below ceiling — no Phase 4 prune
+    expect(result.factsPruned).toBe(0);
+  });
 });
 
 // ---------------------------------------------------------------------------

src/memory/jme.ts

@@ -44,6 +44,14 @@ export const TEMPORAL_DEDUP_THRESHOLD = 0.85;
  */
 export const VECTOR_CEILING_WARN = 400;
 
+/**
+ * Phase 4: When the vector ceiling is active, facts below this confidence
+ * threshold are pruned during consolidation — more aggressive than the
+ * baseline 0.4 floor used in pruneExpiredFacts(). Clears marginal facts
+ * before they degrade recall quality by filling the LIMIT 500 query window.
+ */
+export const PHASE4_CONFIDENCE_FLOOR = 0.5;
+
 // ── Types ────────────────────────────────────────────────────────────────────
 
 export type JmeTurnRole = "user" | "jarvis";
@@ -723,6 +731,8 @@ export interface ConsolidateResult {
   factsInserted: number;
   factsSkipped: number;
   factsSuperseded: number;
+  /** Phase 4: facts pruned during auto-prune when ceiling was active. */
+  factsPruned: number;
 }
 
 /**
@@ -746,6 +756,7 @@ export async function consolidateAll(): Promise<ConsolidateResult> {
     factsInserted: 0,
     factsSkipped: 0,
     factsSuperseded: 0,
+    factsPruned: 0,
   };
 
   try {
@@ -767,10 +778,12 @@ export async function consolidateAll(): Promise<ConsolidateResult> {
       role: JmeTurnRole;
       content: string;
     }>;
-    // Phase 3 Pieza 2: vector ceiling surveillance.
+    // Phase 3 Pieza 2 + Phase 4: vector ceiling surveillance + auto-prune.
     // When jme_facts with embeddings approaches the LIMIT 500 in queryMemory(),
-    // recall quality degrades (oldest facts get cut off). Warn early at 400 so
-    // there's runway before it becomes a problem. Phase 4 will add auto-pruning.
+    // recall quality degrades (oldest facts get cut off). Phase 4: when the
+    // ceiling is active, auto-prune low-confidence facts (threshold raised to
+    // PHASE4_CONFIDENCE_FLOOR=0.5) before extraction, so the query window has
+    // room for the facts that matter.
     const embeddingCount = (
       db
         .prepare(
@@ -780,8 +793,24 @@ export async function consolidateAll(): Promise<ConsolidateResult> {
     ).n;
     if (embeddingCount >= VECTOR_CEILING_WARN) {
       console.warn(
-        `[jme] consolidateAll: vector ceiling warning — ${embeddingCount} facts with embeddings (warn threshold=${VECTOR_CEILING_WARN}, query LIMIT=500). Consider pruning low-confidence facts.`,
+        `[jme] consolidateAll: vector ceiling active — ${embeddingCount} facts (warn=${VECTOR_CEILING_WARN}). Auto-pruning low-confidence facts (threshold=${PHASE4_CONFIDENCE_FLOOR}).`,
       );
+      // Phase 4 auto-prune: delete expired AND low-confidence facts using the
+      // elevated Phase 4 floor. This is a superset of pruneExpiredFacts() —
+      // it catches facts that the baseline 0.4 threshold would have kept.
+      const pruneResult = db
+        .prepare(
+          `DELETE FROM jme_facts
+           WHERE (expires_at IS NOT NULL AND expires_at < ?)
+              OR (confidence < ? AND ts < ?)`,
+        )
+        .run(Date.now(), PHASE4_CONFIDENCE_FLOOR, Date.now() - 30 * 24 * 60 * 60 * 1000);
+      result.factsPruned = pruneResult.changes;
+      if (result.factsPruned > 0) {
+        console.log(
+          `[jme] consolidateAll: phase4 auto-prune removed ${result.factsPruned} low-confidence facts`,
+        );
+      }
     }
 
     if (turns.length === 0) return result;

src/rituals/scheduler.ts

@@ -910,8 +910,9 @@ function scheduleJmeConsolidation(): void {
           await import("../memory/jme.js");
         const result = await consolidateAll();
         const pruned = pruneStaleTurns();
+        const p4Tag = result.factsPruned > 0 ? `, ${result.factsPruned} p4-pruned` : "";
         console.log(
-          `[rituals] jme-consolidate: ${result.turnsProcessed} turns → ${result.factsExtracted} facts (${result.factsInserted} ins, ${result.factsSkipped} skip, ${result.factsSuperseded} sup), ${pruned} stale pruned`,
+          `[rituals] jme-consolidate: ${result.turnsProcessed} turns → ${result.factsExtracted} facts (${result.factsInserted} ins, ${result.factsSkipped} skip, ${result.factsSuperseded} sup), ${pruned} stale pruned${p4Tag}`,
         );
       } catch (err) {
         console.error("[rituals] jme-consolidate failed:", err);