@konneal/engine 0.2.24 → 0.2.25

package/dist/{ask-OO6UUIFF.js → ask-YE2L3Q5H.js}

@@ -1,7 +1,7 @@
 import {
   handleAsk
-} from "./chunk-IM5KMIL6.js";
-import "./chunk-HXWDVAVN.js";
+} from "./chunk-BWSH7VBG.js";
+import "./chunk-6MHR5JI4.js";
 import "./chunk-FYGJDJPN.js";
 import "./chunk-RZ3UZVRR.js";
 import "./chunk-IOZAZA7Y.js";

package/dist/{chunk-HXWDVAVN.js → chunk-6MHR5JI4.js}

@@ -23275,7 +23275,7 @@ async function retrieve(env, query, opts = {}) {
   let rq = opts.queryOverride?.trim() || u?.standalone_query?.trim() || folded;
   if (u?.process_intent) rq += processExpansion();
   const vectorP = rq === folded && opts.optimisticVec ? Promise.resolve(opts.optimisticVec) : rq === folded && opts.warmEmbed ? opts.warmEmbed.then((w) => w ?? embed(portModelRunner(env), MODELS.embed, rq)) : embed(portModelRunner(env), MODELS.embed, rq);
-  const lexicalP = lexicalPrefilter(env, rq).catch(() => []);
+  const lexicalP = lexicalPrefilter(env, opts.lexicalBoost ? `${rq} ${opts.lexicalBoost}` : rq).catch(() => []);
   const [vector, lexicalHits0] = await Promise.all([vectorP, lexicalP]);
   const lexicalHits = opts.sealScope || opts.standardKeys ? lexicalHits0.filter(
     (h) => (!opts.sealScope || h.metadata.doc_number === opts.sealScope.doc_number && (!opts.sealScope.edition || h.metadata.edition === opts.sealScope.edition)) && standardKeyAllowed(h.metadata, opts.standardKeys)

package/dist/{chunk-IM5KMIL6.js → chunk-BWSH7VBG.js}

@@ -27,7 +27,7 @@ import {
   syntheticUnderstanding,
   telemetry,
   understandQuery
-} from "./chunk-HXWDVAVN.js";
+} from "./chunk-6MHR5JI4.js";
 import {
   corsHeaders,
   err,
@@ -2013,10 +2013,12 @@ ${summary}` }] : [],
   } : null;
   if (aggregationVerdict) console.log("aggregation engine:", aggregationVerdict.operation, aggregationVerdict.table, "\u2192", aggregationVerdict.value);
   let boundaryNote = null;
+  let boundaryBoost;
   if (P().sources?.licensed?.length) {
     const match = matchLicensedTopic(q.query, P().sources.licensed);
     if (match && !(standardKeys?.has(match.entry.key) ?? false)) {
       const docNum = match.entry.doc_number ?? "";
+      boundaryBoost = docNum || void 0;
       let citing = [];
       if (docNum) {
         const rows = await env.DB.prepare(
@@ -2062,7 +2064,8 @@ Answer account questions from these records ONLY: name the record when you use i
       optimisticHits,
       optimisticVec,
       datasetScope: narrowed ? corpora : null,
-      standardKeys
+      standardKeys,
+      lexicalBoost: boundaryBoost
     });
     stageTiming["retrieve-core"] = Date.now() - tR;
     console.log("stage: retrieve", Date.now() - tR, "ms");
@@ -2086,7 +2089,7 @@ Answer account questions from these records ONLY: name the record when you use i
     if (grade === "weak" && understanding?.docidentifier) {
       const broaden = `${understanding.standalone_query || q.query} ${understanding.docidentifier}`.trim();
       const tc = Date.now();
-      const second = await retrieve(env, q.query, { prev, understanding, queryOverride: broaden, federate, datasetScope: narrowed ? corpora : null, standardKeys, sealScope: declaredScoped ? docScope : null });
+      const second = await retrieve(env, q.query, { prev, understanding, queryOverride: broaden, federate, datasetScope: narrowed ? corpora : null, standardKeys, sealScope: declaredScoped ? docScope : null, lexicalBoost: boundaryBoost });
       const grade2 = await gradeRetrieval(env.AI, roleModel(env, "grader"), q.query, second.hits.map((h) => h.text));
       stageTiming.corrective = Date.now() - tc;
       if (grade2 === "good") retrieved = second;

package/dist/{chunk-PZLYPUYJ.js → chunk-WQ2PYIWK.js}

@@ -7,7 +7,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "./chunk-HXWDVAVN.js";
+} from "./chunk-6MHR5JI4.js";
 import {
   corsHeaders,
   err,

package/dist/{search-OJ776DMH.js → search-I26YUVSP.js}

@@ -1,7 +1,7 @@
 import {
   handleSearch
-} from "./chunk-PZLYPUYJ.js";
-import "./chunk-HXWDVAVN.js";
+} from "./chunk-WQ2PYIWK.js";
+import "./chunk-6MHR5JI4.js";
 import "./chunk-FYGJDJPN.js";
 import "./chunk-RZ3UZVRR.js";
 import "./chunk-IOZAZA7Y.js";

package/dist/worker_public/src/index.js

@@ -1,13 +1,13 @@
 import {
   handleSearch
-} from "../../chunk-PZLYPUYJ.js";
+} from "../../chunk-WQ2PYIWK.js";
 import {
   checkQuoteAnchors,
   handleAsk,
   handleMemories,
   scoreJudge,
   standardForDocNumber
-} from "../../chunk-IM5KMIL6.js";
+} from "../../chunk-BWSH7VBG.js";
 import {
   buildMessages,
   citations,
@@ -29,7 +29,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "../../chunk-HXWDVAVN.js";
+} from "../../chunk-6MHR5JI4.js";
 import {
   authenticate,
   corsHeaders,
@@ -979,7 +979,7 @@ async function handleMcp(env, ctx, req, tier, key) {
       // stream:false forces the JSON lane (anon defaults to SSE)
       body: JSON.stringify({ ...args, stream: false })
     });
-    const res = name === "ask" ? await (await import("../../ask-OO6UUIFF.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-OJ776DMH.js")).handleSearch(env, ctx, inner, tier, key);
+    const res = name === "ask" ? await (await import("../../ask-YE2L3Q5H.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-I26YUVSP.js")).handleSearch(env, ctx, inner, tier, key);
     return res.json().catch(() => ({ error: { message: "tool transport failed", status: res.status } }));
   });
   if (out.ok && "accepted" in out) return new Response(null, { status: 202 });

package/package.json

@@ -27,7 +27,7 @@
     "playwright": "^1.62.1",
     "yaml": "^2.9.1"
   },
-  "version": "0.2.24",
+  "version": "0.2.25",
   "description": "The Konneal engine: the publisher-agnostic build pipeline and API plane for standards intelligence (retrieval, answer contract, verdicts, evaluation).",
   "license": "BSD-3-Clause",
   "type": "module",

package/workers/worker_public/src/ask.ts

@@ -853,10 +853,16 @@ async function handleAsk(
   // reference the licensed document, and the note instructs the model
   // to attribute, never to recite the licensed parameters.
   let boundaryNote: string | null = null;
+  // the boost rides RETRIEVAL (the lexical lane below): chunks whose
+  // prose references the licensed document — the referencing
+  // publication's own clause — surface for a question the document's
+  // vocabulary alone would miss
+  let boundaryBoost: string | undefined;
   if (P().sources?.licensed?.length) {
     const match = matchLicensedTopic(q.query, P().sources.licensed);
     if (match && !(standardKeys?.has(match.entry.key) ?? false)) {
       const docNum = match.entry.doc_number ?? "";
+      boundaryBoost = docNum || undefined;
       let citing: string[] = [];
       if (docNum) {
         const rows = await env.DB.prepare(
@@ -923,7 +929,7 @@ async function handleAsk(
     // publications there).
     retrieved = await retrieve(env, q.query, { prev, understanding, federate, warmEmbed, graphDocNumbers,
       sealScope: declaredScoped ? docScope : null, optimisticHits, optimisticVec,
-      datasetScope: narrowed ? corpora : null, standardKeys });
+      datasetScope: narrowed ? corpora : null, standardKeys, lexicalBoost: boundaryBoost });
     stageTiming["retrieve-core"] = Date.now() - tR;
     console.log("stage: retrieve", Date.now() - tR, "ms");
     // ── TTFT surgery: the two post-retrieval LLM calls run IN PARALLEL —
@@ -957,7 +963,7 @@ async function handleAsk(
     if (grade === "weak" && understanding?.docidentifier) {
       const broaden = `${understanding.standalone_query || q.query} ${understanding.docidentifier}`.trim();
       const tc = Date.now();
-      const second = await retrieve(env, q.query, { prev, understanding, queryOverride: broaden, federate, datasetScope: narrowed ? corpora : null, standardKeys, sealScope: declaredScoped ? docScope : null });
+      const second = await retrieve(env, q.query, { prev, understanding, queryOverride: broaden, federate, datasetScope: narrowed ? corpora : null, standardKeys, sealScope: declaredScoped ? docScope : null, lexicalBoost: boundaryBoost });
       const grade2 = await gradeRetrieval(env.AI, roleModel(env, "grader"), q.query, second.hits.map((h: Hit) => h.text));
       stageTiming.corrective = Date.now() - tc;
       if (grade2 === "good") retrieved = second; // corrective retry must be strictly better

package/workers/worker_public/src/pipeline.ts

@@ -120,7 +120,7 @@ export async function retrieve(
       : rq === folded && opts.warmEmbed
         ? opts.warmEmbed.then((w) => w ?? embed(portModelRunner(env), MODELS.embed, rq))
         : embed(portModelRunner(env), MODELS.embed, rq);
-  const lexicalP = lexicalPrefilter(env, rq).catch(() => [] as Hit[]);
+  const lexicalP = lexicalPrefilter(env, opts.lexicalBoost ? `${rq} ${opts.lexicalBoost}` : rq).catch(() => [] as Hit[]);
   const [vector, lexicalHits0] = await Promise.all([vectorP, lexicalP]);
   // The declared context's seal binds the lexical lane at the SOURCE: the
   // RRF fusion mixes the full-corpus lexical ranking straight into the

package/workers/worker_public/src/stages/types.ts

@@ -12,6 +12,11 @@ export interface RetrieveOptions {
   prev?: string;
   understanding?: QueryUnderstanding | null;
   queryOverride?: string;
+  /** Extra terms appended to the LEXICAL lane's query only (never the
+   *  dense vector): the licensed-boundary boost — chunks whose prose
+   *  references the licensed document surface for a question the
+   *  document's own vocabulary would otherwise miss. */
+  lexicalBoost?: string;
   federate?: (query: string) => Promise<Hit[]>;
   warmEmbed?: Promise<number[] | null>;
   graphDocNumbers?: string[];