feat(threshold): raise default private_date post-filter to 0.85

src/defaults.ts

@@ -8,7 +8,7 @@
 // Strict TS index-access fallbacks (e.g. `arr[i] ?? 0` under
 // `noUncheckedIndexedAccess`) are NOT user-facing defaults and stay inline.
 
-import type { BackendName, ModelVariant } from './types/index.js';
+import type { BackendName, ModelVariant, PiiCategory } from './types/index.js';
 import type { Recognizer } from './types/recognizer.js';
 import {
   base58CheckValid,
@@ -833,6 +833,19 @@ export const DEFAULT_DECODE_THRESHOLD = 0.5;
  * Override via `NullPiiConfig.threshold` / `categoryThresholds`. */
 export const DEFAULT_POST_FILTER_THRESHOLD = 0;
 
+/** Per-category post-filter thresholds layered on top of the global
+ * default. User-supplied `categoryThresholds` win on a key-by-key
+ * basis (only the labels the user spells out are overridden).
+ *
+ * `private_date` — model is over-eager on calendar dates without
+ * identifying context (e.g. "Today's date is 2026-05-21" in a system
+ * prompt or footer). Tightening to 0.85 drops the long tail of
+ * low-confidence date hits while keeping DOB / birth-date / explicit
+ * date PII (which the model emits well above 0.9). */
+export const DEFAULT_CATEGORY_THRESHOLDS: Partial<Record<PiiCategory, number>> = {
+  private_date: 0.85,
+};
+
 /** IoU threshold used by `dedupeOverlappingSpans` when reconciling
  * ML + recognizer spans. Two spans with IoU at or above this are
  * treated as duplicates; higher-scoring one wins. */

src/nullpii.ts

@@ -7,6 +7,7 @@ import { NULLPII_MODEL_DIR, readEnvVar } from './config.js';
 import {
   BOUNDARY_REFINE_TRIM_CHARS,
   DEFAULT_BOUNDARY_REFINE,
+  DEFAULT_CATEGORY_THRESHOLDS,
   DEFAULT_DECODE_THRESHOLD,
   DEFAULT_DEDUPE_IOU,
   DEFAULT_POST_FILTER_THRESHOLD,
@@ -225,10 +226,18 @@ export class NullPii {
       DEFAULT_DEDUPE_IOU,
       { acrossLabels: true },
     ) as PiiSpan[];
+    // Merge built-in per-category defaults with user overrides — user
+    // keys win on a key-by-key basis. The built-in `private_date: 0.85`
+    // drops the long tail of low-confidence calendar-date hits the model
+    // emits without identifying context.
+    const mergedCategoryThresholds = {
+      ...DEFAULT_CATEGORY_THRESHOLDS,
+      ...(this.config.categoryThresholds ?? {}),
+    };
     const merged = applyThresholds(
       combined,
       this.config.threshold ?? DEFAULT_POST_FILTER_THRESHOLD,
-      this.config.categoryThresholds ?? {},
+      mergedCategoryThresholds,
     );
     const refineOn = this.config.boundaryRefine ?? DEFAULT_BOUNDARY_REFINE;
     const spans = refineOn ? refineSpanBoundaries(escaped, merged) : merged;

test/nullpii.test.ts

@@ -198,6 +198,55 @@ describe('NullPii e2e pipeline (mocked ONNX)', () => {
     await n.dispose();
   });
 
+  it('drops low-confidence private_date spans via built-in category threshold', async () => {
+    // Default `categoryThresholds.private_date` = 0.85 — a recognizer
+    // that fires at 0.7 must be dropped; one at 0.9 must survive. This
+    // protects against the gateway-report case of innocuous calendar
+    // dates in system prompts being placeholdered.
+    const n = new NullPii({
+      modelDir: '/fake',
+      backend: 'cpu',
+      recognizers: [
+        {
+          id: 'test:weak-date',
+          pattern: /\b2026-05-21\b/g,
+          label: 'private_date',
+          confidence: 0.7,
+        },
+        {
+          id: 'test:strong-date',
+          pattern: /\b1985-03-12\b/g,
+          label: 'private_date',
+          confidence: 0.9,
+        },
+      ],
+    });
+    const out = await n.sanitize('Today 2026-05-21 — DOB 1985-03-12');
+    expect(out.spans).toHaveLength(1);
+    expect(out.spans[0]?.text).toBe('1985-03-12');
+    await n.dispose();
+  });
+
+  it('user categoryThresholds override built-in date default per-key', async () => {
+    // Caller can lower the bar back down for their own use case.
+    const n = new NullPii({
+      modelDir: '/fake',
+      backend: 'cpu',
+      categoryThresholds: { private_date: 0.5 },
+      recognizers: [
+        {
+          id: 'test:weak-date',
+          pattern: /\b2026-05-21\b/g,
+          label: 'private_date',
+          confidence: 0.7,
+        },
+      ],
+    });
+    const out = await n.sanitize('Today 2026-05-21');
+    expect(out.spans).toHaveLength(1);
+    await n.dispose();
+  });
+
   it('init runs once across many sanitize calls', async () => {
     const n = new NullPii({ modelDir: '/fake', backend: 'cpu' });
     const r1 = await n.sanitize('First email: a@acme.io');