audrey 1.0.1 → 1.0.3

package/benchmarks/output/submission-bundle/submission-manifest.json

@@ -1,7 +1,7 @@
 {
   "schemaVersion": "1.0.0",
   "suite": "GuardBench submission bundle",
-  "generatedAt": "2026-05-15T17:52:13.050Z",
+  "generatedAt": "2026-05-29T13:33:23.534Z",
   "sourceDir": "benchmarks/output",
   "subject": {
     "name": "Audrey Guard",
@@ -16,9 +16,9 @@
     "evidenceRecall": 1,
     "redactionLeaks": 0,
     "latency": {
-      "p50Ms": 2.465,
-      "p95Ms": 30.791,
-      "maxMs": 30.791
+      "p50Ms": 3.09,
+      "p95Ms": 28.181,
+      "maxMs": 28.181
     }
   },
   "conformance": {
@@ -44,8 +44,8 @@
   "files": [
     {
       "path": "guardbench-conformance-card.json",
-      "bytes": 1735,
-      "sha256": "65d4a10afe10178e4e4e3d508e0e06ce5a4b28713debc27cf4d9b60a46050d2f"
+      "bytes": 1748,
+      "sha256": "88cba808b893cde3530f4e29317d4cf9ae3b20f08456c58d97d4e5300f5f9298"
     },
     {
       "path": "guardbench-manifest.json",
@@ -54,13 +54,13 @@
     },
     {
       "path": "guardbench-raw.json",
-      "bytes": 43455,
-      "sha256": "3b78d1a2432e7d72752f96d9ac4b2b49cf6f59eb65548fbadb21ea6adbb86b37"
+      "bytes": 43466,
+      "sha256": "66d4b69087258638f3572a40e1fd59bb84067034f899eaa2c27eed2dde554b2b"
     },
     {
       "path": "guardbench-summary.json",
-      "bytes": 69350,
-      "sha256": "21023f230b761f1b43f8ecabe519dd6b320c62ad56f0b6aa28bbcf7a2c8838f5"
+      "bytes": 69359,
+      "sha256": "91f264dd889e2c639a6fc6d1b867bc228b94c84ed5120345e23dddb79c11ee74"
     },
     {
       "path": "schemas/guardbench-adapter-registry.schema.json",
@@ -125,7 +125,7 @@
     {
       "path": "validation-report.json",
       "bytes": 739,
-      "sha256": "570d28760ad3611ccf4f08e98281a3fe659730463d13e96ff31073c7ee10ce65"
+      "sha256": "284c40f608508d59e8cdc8047371dfcf0a9824a891ff37b77035241ab5de275f"
     }
   ]
 }

package/benchmarks/output/submission-bundle/validation-report.json

@@ -1,5 +1,5 @@
 {
-  "generatedAt": "2026-05-15T17:52:13.050Z",
+  "generatedAt": "2026-05-29T13:33:23.534Z",
   "sourceValidation": {
     "ok": true,
     "dir": "benchmarks/output",

package/benchmarks/output/summary.json

@@ -1,5 +1,5 @@
 {
-  "generatedAt": "2026-05-15T17:53:31.952Z",
+  "generatedAt": "2026-05-29T13:34:04.184Z",
   "command": "node benchmarks/run.js --provider mock --dimensions 64",
   "config": {
     "provider": "mock",
@@ -23,19 +23,19 @@
         "system": "Audrey",
         "scorePercent": 100,
         "passRate": 100,
-        "avgDurationMs": 26.333333333333332
+        "avgDurationMs": 14.166666666666666
       },
       {
         "system": "Vector Only",
         "scorePercent": 41.66666666666667,
         "passRate": 25,
-        "avgDurationMs": 0.3333333333333333
+        "avgDurationMs": 0.5
       },
       {
         "system": "Keyword + Recency",
         "scorePercent": 41.66666666666667,
         "passRate": 25,
-        "avgDurationMs": 0.5
+        "avgDurationMs": 0.5833333333333334
       },
       {
         "system": "Recent Window",
@@ -170,7 +170,7 @@
             "system": "Audrey",
             "scorePercent": 100,
             "passRate": 100,
-            "avgDurationMs": 24
+            "avgDurationMs": 14.625
           },
           {
             "system": "Vector Only",
@@ -182,7 +182,7 @@
             "system": "Keyword + Recency",
             "scorePercent": 50,
             "passRate": 37.5,
-            "avgDurationMs": 0.75
+            "avgDurationMs": 0.875
           },
           {
             "system": "Recent Window",
@@ -277,7 +277,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 34,
+                "durationMs": 23,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -289,7 +289,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 1,
+                "durationMs": 2,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -301,7 +301,7 @@
               },
               {
                 "system": "Keyword + Recency",
-                "durationMs": 6,
+                "durationMs": 7,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -336,7 +336,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 20,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -347,7 +347,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 0,
+                "durationMs": 1,
                 "passed": false,
                 "score": 0.5,
                 "summary": "retrieved expected evidence, but conflicting evidence still appeared later",
@@ -394,7 +394,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 25,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -407,7 +407,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 1,
+                "durationMs": 0,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -457,7 +457,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 19,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -518,7 +518,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 20,
+                "durationMs": 12,
                 "passed": true,
                 "score": 1,
                 "summary": "correct abstention",
@@ -574,7 +574,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 38,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -632,7 +632,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 22,
+                "durationMs": 17,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -693,7 +693,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 14,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "correct abstention",
@@ -702,7 +702,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 1,
+                "durationMs": 0,
                 "passed": false,
                 "score": 0,
                 "summary": "leaked restricted content",
@@ -750,7 +750,7 @@
             "system": "Audrey",
             "scorePercent": 100,
             "passRate": 100,
-            "avgDurationMs": 31
+            "avgDurationMs": 13.25
           },
           {
             "system": "Keyword + Recency",
@@ -762,7 +762,7 @@
             "system": "Vector Only",
             "scorePercent": 12.5,
             "passRate": 0,
-            "avgDurationMs": 0.25
+            "avgDurationMs": 0.75
           },
           {
             "system": "Recent Window",
@@ -821,7 +821,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 20,
+                "durationMs": 11,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -879,7 +879,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 19,
+                "durationMs": 12,
                 "passed": true,
                 "score": 1,
                 "summary": "correct abstention",
@@ -888,7 +888,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 0,
+                "durationMs": 1,
                 "passed": false,
                 "score": 0,
                 "summary": "leaked restricted content",
@@ -932,7 +932,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 63,
+                "durationMs": 14,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -993,7 +993,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 22,
+                "durationMs": 16,
                 "passed": true,
                 "score": 1,
                 "summary": "retrieved expected evidence",
@@ -1004,7 +1004,7 @@
               },
               {
                 "system": "Vector Only",
-                "durationMs": 0,
+                "durationMs": 1,
                 "passed": false,
                 "score": 0,
                 "summary": "missed target evidence",
@@ -1118,7 +1118,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 15,
+                "durationMs": 18,
                 "passed": true,
                 "score": 1,
                 "summary": "matched all required signals",
@@ -1176,7 +1176,7 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 14,
+                "durationMs": 12,
                 "passed": true,
                 "score": 1,
                 "summary": "matched all required signals",
@@ -1234,12 +1234,12 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 15,
+                "durationMs": 13,
                 "passed": true,
                 "score": 1,
                 "summary": "matched all required signals",
                 "topResults": [
-                  "guard_hardened:replay_rejected error:guard receipt already has an outcome: 01KRPCBA92C019YZAVN2M187H1",
+                  "guard_hardened:replay_rejected error:guard receipt already has an outcome: 01KSSZ29C75TB1E56B31P9ZW4B",
                   "decision:caution verdict:caution risk:0.55 Caution: 1 memory signal, 1 medium severity found before acting.",
                   "warning:recent_failure severity:medium npm test failed 1x recently: Vitest failed with spawn EPERM Before re-running npm test, check what changed since the last failure."
                 ],
@@ -1292,12 +1292,12 @@
             "results": [
               {
                 "system": "Audrey",
-                "durationMs": 10,
+                "durationMs": 11,
                 "passed": true,
                 "score": 1,
                 "summary": "matched all required signals",
                 "topResults": [
-                  "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01KRPCBA9CMSC7CEQJPFZA6E17",
+                  "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01KSSZ29CKZF534VNBKZV6WGSP",
                   "decision:go verdict:clear risk:0 No relevant memory risks, prior failures, or must-follow procedures were found."
                 ],
                 "retrievalSummary": "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01K | decision:go verdict:clear risk:0 No relevant memory risks, prior failure"
@@ -1352,7 +1352,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 34,
+            "durationMs": 23,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1364,7 +1364,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 1,
+            "durationMs": 2,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1376,7 +1376,7 @@
           },
           {
             "system": "Keyword + Recency",
-            "durationMs": 6,
+            "durationMs": 7,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1411,7 +1411,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 20,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1422,7 +1422,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 0,
+            "durationMs": 1,
             "passed": false,
             "score": 0.5,
             "summary": "retrieved expected evidence, but conflicting evidence still appeared later",
@@ -1469,7 +1469,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 25,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1482,7 +1482,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 1,
+            "durationMs": 0,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1532,7 +1532,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 19,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1593,7 +1593,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 20,
+            "durationMs": 12,
             "passed": true,
             "score": 1,
             "summary": "correct abstention",
@@ -1649,7 +1649,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 38,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1707,7 +1707,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 22,
+            "durationMs": 17,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1768,7 +1768,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 14,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "correct abstention",
@@ -1777,7 +1777,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 1,
+            "durationMs": 0,
             "passed": false,
             "score": 0,
             "summary": "leaked restricted content",
@@ -1824,7 +1824,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 20,
+            "durationMs": 11,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1882,7 +1882,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 19,
+            "durationMs": 12,
             "passed": true,
             "score": 1,
             "summary": "correct abstention",
@@ -1891,7 +1891,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 0,
+            "durationMs": 1,
             "passed": false,
             "score": 0,
             "summary": "leaked restricted content",
@@ -1935,7 +1935,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 63,
+            "durationMs": 14,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -1996,7 +1996,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 22,
+            "durationMs": 16,
             "passed": true,
             "score": 1,
             "summary": "retrieved expected evidence",
@@ -2007,7 +2007,7 @@
           },
           {
             "system": "Vector Only",
-            "durationMs": 0,
+            "durationMs": 1,
             "passed": false,
             "score": 0,
             "summary": "missed target evidence",
@@ -2058,7 +2058,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 15,
+            "durationMs": 18,
             "passed": true,
             "score": 1,
             "summary": "matched all required signals",
@@ -2116,7 +2116,7 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 14,
+            "durationMs": 12,
             "passed": true,
             "score": 1,
             "summary": "matched all required signals",
@@ -2174,12 +2174,12 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 15,
+            "durationMs": 13,
             "passed": true,
             "score": 1,
             "summary": "matched all required signals",
             "topResults": [
-              "guard_hardened:replay_rejected error:guard receipt already has an outcome: 01KRPCBA92C019YZAVN2M187H1",
+              "guard_hardened:replay_rejected error:guard receipt already has an outcome: 01KSSZ29C75TB1E56B31P9ZW4B",
               "decision:caution verdict:caution risk:0.55 Caution: 1 memory signal, 1 medium severity found before acting.",
               "warning:recent_failure severity:medium npm test failed 1x recently: Vitest failed with spawn EPERM Before re-running npm test, check what changed since the last failure."
             ],
@@ -2232,12 +2232,12 @@
         "results": [
           {
             "system": "Audrey",
-            "durationMs": 10,
+            "durationMs": 11,
             "passed": true,
             "score": 1,
             "summary": "matched all required signals",
             "topResults": [
-              "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01KRPCBA9CMSC7CEQJPFZA6E17",
+              "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01KSSZ29CKZF534VNBKZV6WGSP",
               "decision:go verdict:clear risk:0 No relevant memory risks, prior failures, or must-follow procedures were found."
             ],
             "retrievalSummary": "guard_hardened:non_guard_receipt_rejected error:not a guard receipt: 01K | decision:go verdict:clear risk:0 No relevant memory risks, prior failure"

package/benchmarks/perf-snapshot.js

@@ -157,8 +157,8 @@ function parseArgs(argv = process.argv.slice(2)) {
     if (token === '--sizes' && argv[i + 1]) {
       args.sizes = argv[++i]
         .split(',')
-        .map((s) => Number.parseInt(s.trim(), 10))
-        .filter((n) => Number.isFinite(n) && n > 0);
+        .map(s => Number.parseInt(s.trim(), 10))
+        .filter(n => Number.isFinite(n) && n > 0);
     } else if (token === '--recall-runs' && argv[i + 1]) {
       args.recallRuns = Number.parseInt(argv[++i], 10);
     } else if (token === '--out' && argv[i + 1]) {
@@ -180,7 +180,7 @@ async function runOneSize({ size, recallRuns }) {
   });
 
   const queueProcessingTimes = [];
-  audrey.on('post-encode-complete', (event) => {
+  audrey.on('post-encode-complete', event => {
     queueProcessingTimes.push(event.processing_ms);
   });
 
@@ -223,7 +223,10 @@ async function runOneSize({ size, recallRuns }) {
   }
 }
 
-export async function runPerfSnapshot({ sizes = DEFAULT_SIZES, recallRuns = DEFAULT_RECALL_RUNS } = {}) {
+export async function runPerfSnapshot({
+  sizes = DEFAULT_SIZES,
+  recallRuns = DEFAULT_RECALL_RUNS,
+} = {}) {
   const startedAt = Date.now();
   const sized = [];
   for (const size of sizes) {
@@ -265,11 +268,11 @@ export function formatMarkdownTable(snapshot) {
   lines.push(
     `Node ${snapshot.machine.node} · ${snapshot.machine.cpuCount}x ${snapshot.machine.cpuModel} · ${snapshot.machine.memoryGb} GB RAM`,
   );
+  lines.push(`Generated ${snapshot.generatedAt}${snapshot.gitSha ? ` (${snapshot.gitSha})` : ''}`);
+  lines.push('');
   lines.push(
-    `Generated ${snapshot.generatedAt}${snapshot.gitSha ? ` (${snapshot.gitSha})` : ''}`,
+    '| Corpus size | Encode p50 (ms) | Encode p95 (ms) | Recall p50 (ms) | Recall p95 (ms) | Recall p99 (ms) |',
   );
-  lines.push('');
-  lines.push('| Corpus size | Encode p50 (ms) | Encode p95 (ms) | Recall p50 (ms) | Recall p95 (ms) | Recall p99 (ms) |');
   lines.push('|---|---|---|---|---|---|');
   for (const row of snapshot.sizes) {
     lines.push(
@@ -287,7 +290,7 @@ export function formatMarkdownTable(snapshot) {
 if (process.argv[1] && import.meta.url === pathToFileURL(process.argv[1]).href) {
   const args = parseArgs();
   runPerfSnapshot({ sizes: args.sizes, recallRuns: args.recallRuns })
-    .then((snapshot) => {
+    .then(snapshot => {
       if (args.out) {
         writeFileSync(args.out, JSON.stringify(snapshot, null, 2) + '\n');
       }
@@ -297,7 +300,7 @@ if (process.argv[1] && import.meta.url === pathToFileURL(process.argv[1]).href)
         process.stdout.write(formatMarkdownTable(snapshot) + '\n');
       }
     })
-    .catch((err) => {
+    .catch(err => {
       console.error('[audrey] perf snapshot failed:', err);
       process.exit(1);
     });

package/benchmarks/perf.bench.js

@@ -136,16 +136,24 @@ export async function runPerfBenchmark({
     };
 
     if (queueProcessingTimes.length !== runs) {
-      throw new Error(`expected ${runs} post-encode queue events, got ${queueProcessingTimes.length}`);
+      throw new Error(
+        `expected ${runs} post-encode queue events, got ${queueProcessingTimes.length}`,
+      );
     }
 
     assertBudget('encode response p95', result.encode_response_ms.p95, budgets.encodeResponseP95Ms);
     assertBudget('hybrid recall p95', result.hybrid_recall_ms.p95, budgets.hybridRecallP95Ms);
-    assertBudget('queue processing p50', result.queue_processing_ms.p50, budgets.queueProcessingP50Ms);
-
-    out(`Audrey perf gate passed: encode p95=${result.encode_response_ms.p95}ms, `
-      + `hybrid recall p95=${result.hybrid_recall_ms.p95}ms, `
-      + `queue p50=${result.queue_processing_ms.p50}ms`);
+    assertBudget(
+      'queue processing p50',
+      result.queue_processing_ms.p50,
+      budgets.queueProcessingP50Ms,
+    );
+
+    out(
+      `Audrey perf gate passed: encode p95=${result.encode_response_ms.p95}ms, ` +
+        `hybrid recall p95=${result.hybrid_recall_ms.p95}ms, ` +
+        `queue p50=${result.queue_processing_ms.p50}ms`,
+    );
     return result;
   } finally {
     audrey.close();