@konneal/engine 0.2.100 → 0.2.102

package/dist/{ask-KP42T6DH.js → ask-E6YJUOIP.js}

@@ -1,8 +1,8 @@
 import {
   handleAsk
-} from "./chunk-GYCUMFOA.js";
+} from "./chunk-FQVP3JUA.js";
 import "./chunk-QNLFNEI4.js";
-import "./chunk-BBKAETKM.js";
+import "./chunk-25ECO5M7.js";
 import "./chunk-KNCRYAX3.js";
 import "./chunk-DTW4UW2E.js";
 import "./chunk-A3QHHUN5.js";

package/dist/{chunk-BBKAETKM.js → chunk-25ECO5M7.js}

@@ -22318,27 +22318,27 @@ async function runStages(stages, c) {
 var dense = {
   name: "dense",
   run: async (c) => {
-    const { env, filter, filters, vector, opts, rq, folded } = c;
+    const { env, filters, vector, opts, rq, folded } = c;
     const q = { topK: LIMITS.retrieveK, returnMetadata: "all" };
-    if (filter) q.filter = filter;
     const optimistic = opts.optimisticHits ?? [];
     const sameLane = rq === folded;
-    if (!filter && sameLane && optimistic.length) {
+    if (!filters?.doc_number && sameLane && optimistic.length) {
       c.matches = optimistic.map((h) => ({ id: h.id, score: h.score, metadata: h.metadata }));
       console.log("optimistic lane: reused", c.matches.length, "dense hits (no re-query)");
       return;
     }
-    if (filter) {
-      let matches = (await env.VECTORIZE.query(vector, q)).matches ?? [];
+    if (filters?.doc_number) {
+      const allMatches = (await env.VECTORIZE.query(vector, q)).matches ?? [];
+      let matches = allMatches;
+      const before = matches.length;
+      matches = matches.filter((m) => String(m.metadata?.doc_number ?? "") === filters.doc_number);
+      if (filters.edition) matches = matches.filter((m) => !m.metadata?.edition || String(m.metadata.edition) === filters.edition);
+      console.log("dense scope (in code):", before, "\u2192", matches.length, "in doc", filters.doc_number, filters.edition ? `@${filters.edition}` : "");
       if (filters && filters.edition && matches.length < 3) {
-        const docOnly = await env.VECTORIZE.query(vector, {
-          topK: LIMITS.retrieveK,
-          returnMetadata: "all",
-          filter: toVectorizeFilter({ doc_number: filters.doc_number })
-        });
-        if ((docOnly.matches ?? []).length > matches.length) {
-          console.log("edition pin dropped:", filters.doc_number, "@", filters.edition, "\u2192", docOnly.matches?.length ?? 0, "doc-scoped hits (edition not in corpus)");
-          matches = docOnly.matches ?? [];
+        const docOnly = allMatches.filter((m) => String(m.metadata?.doc_number ?? "") === filters.doc_number);
+        if (docOnly.length > matches.length) {
+          console.log("edition pin dropped:", filters.doc_number, "@", filters.edition, "\u2192", docOnly.length, "doc-scoped hits (edition not in corpus)");
+          matches = docOnly;
           filters.edition = void 0;
         }
       }
@@ -22520,10 +22520,10 @@ var conceptGraph = {
       }
     }
     if (numbers.size) {
-      const gc = await lexicalWithin(c.env, c.rq || c.query, [...numbers], 12);
+      const gc = await lexicalWithin(c.env, c.rq || c.query, [...numbers], 6);
       const seenIds0 = new Set(c.matches.map((m) => m.id));
       let merged0 = 0;
-      for (const m of gc.slice(0, 6)) {
+      for (const m of gc.slice(0, 3)) {
         if (!seenIds0.has(m.id)) {
           c.matches.push({ id: m.id, score: m.score * THRESHOLDS.conceptGraphDiscount, metadata: m.metadata });
           seenIds0.add(m.id);
@@ -22547,9 +22547,9 @@ var graphLane = {
     const g = await c.lane["graph-lane"];
     const seenIds = new Set(c.matches.map((m) => m.id));
     let merged = 0;
-    for (const m of (g ?? []).slice(0, 10)) {
+    for (const m of (g ?? []).slice(0, 5)) {
       if (!seenIds.has(m.id)) {
-        c.matches.push({ id: m.id, score: m.score * THRESHOLDS.graphLaneDiscount, metadata: m.metadata });
+        c.matches.push({ id: m.id, score: m.score * THRESHOLDS.graphLaneDiscount * 0.5, metadata: m.metadata });
         seenIds.add(m.id);
         merged++;
       }
@@ -22580,17 +22580,40 @@ var licensedLane = {
     const matches = await portIndex(c.env, "public").getByIds(ids);
     const seenIds = new Set(c.matches.map((m) => m.id));
     let merged = 0;
+    const mergedIds = [];
     for (const m of matches) {
       if (m.metadata && !seenIds.has(m.id)) {
         c.matches.push({ id: m.id, score: THRESHOLDS.licensedLaneScore, metadata: m.metadata });
         seenIds.add(m.id);
         merged++;
+        mergedIds.push(m.id);
       }
     }
+    c.lane["licensed-ids"] = Promise.resolve(mergedIds);
     console.log("licensed lane:", ids.length, "units,", matches.length, "vectors,", merged, "merged");
   }
 };
 
+// workers/worker_public/src/stages/licensedCover.ts
+var licensedCover = {
+  name: "licensed-cover",
+  failure: "additive",
+  when: (c) => !!c.opts.licensedDocNumbers?.length && c.finalHits.length > 0,
+  run: async (c) => {
+    const keys = new Set(c.opts.licensedDocNumbers);
+    const present = c.finalHits.some((h) => keys.has(String(h.metadata.standard ?? "")));
+    if (present) return;
+    const ids = (await c.lane["licensed-ids"] ?? []).slice(0, 8);
+    if (!ids.length) return;
+    const matches = await portIndex(c.env, "public").getByIds(ids);
+    const unit = matches.find((m) => m.metadata && keys.has(String(m.metadata.standard ?? "")));
+    if (!unit) return;
+    const hit = toHits([unit])[0];
+    c.finalHits = [...c.finalHits, hit];
+    console.log("licensed cover: seated", unit.id, "\u2014", unit.metadata.docidentifier);
+  }
+};
+
 // workers/worker_public/src/stages/multiQuery.ts
 var RRF_K = 60;
 var multiQuery = {
@@ -22798,12 +22821,10 @@ var editionCover = {
     let added = 0;
     for (const w of want.slice(0, maxDocs)) {
       try {
-        const q = await c.env.VECTORIZE.query(c.vector, {
-          topK: 3,
-          returnMetadata: "all",
-          filter: { $and: [{ docidentifier: { $eq: w.di } }, { edition: { $eq: w.edition } }] }
-        });
-        const hits = toHits(q.matches ?? []).map((h) => ({ ...h, score: top * THRESHOLDS.editionCoverDiscount }));
+        const rows2 = await c.env.DB.prepare(
+          "SELECT id, doc_id, docidentifier, doctype, doc_number, edition, language, clause_anchor, clause_title, status, superseded_by, corpus, tier, text FROM chunks WHERE docidentifier = ?1 AND edition = ?2 LIMIT 3"
+        ).bind(w.di, w.edition).all();
+        const hits = rowsToHits(rows2.results ?? []).map((h) => ({ ...h, score: top * THRESHOLDS.editionCoverDiscount }));
         c.hits.push(...hits);
         added += hits.length;
         console.log("edition cover:", w.di, w.edition, `+${hits.length}`);
@@ -23223,6 +23244,7 @@ var STAGES = [
   editionSteer,
   propagate,
   diversity,
+  licensedCover,
   typedPin,
   sectionDescent,
   dedup,

package/dist/{chunk-GYCUMFOA.js → chunk-FQVP3JUA.js}

@@ -56,7 +56,7 @@ import {
   tokenBudget,
   understandQuery,
   usageTotal
-} from "./chunk-BBKAETKM.js";
+} from "./chunk-25ECO5M7.js";
 import {
   embed,
   generateOnce,

package/dist/{chunk-OREMF5LK.js → chunk-TFGJ6NYS.js}

@@ -6,7 +6,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "./chunk-BBKAETKM.js";
+} from "./chunk-25ECO5M7.js";
 import {
   portModelRunner
 } from "./chunk-KNCRYAX3.js";

package/dist/lexical.d.ts

@@ -2,6 +2,7 @@ import type { Hit } from "./pipeline";
 /** Build a safe FTS5 MATCH query from user text: alphanumeric tokens,
  *  joined with OR so jargon hits don't require full-phrase match. */
 export declare function ftsMatchQuery(query: string): string | null;
+export declare function rowsToHits(rows: any[]): Hit[];
 /** Lexical ranking WITHIN a document set (the graph lanes' identity
  *  resolution): BM25 over the corpus's own FTS, restricted to the
  *  documents' doc_number values. This is the working substitute for

package/dist/{search-4CBKKNBJ.js → search-5SNP5CHL.js}

@@ -1,7 +1,7 @@
 import {
   handleSearch
-} from "./chunk-OREMF5LK.js";
-import "./chunk-BBKAETKM.js";
+} from "./chunk-TFGJ6NYS.js";
+import "./chunk-25ECO5M7.js";
 import "./chunk-KNCRYAX3.js";
 import "./chunk-DTW4UW2E.js";
 import "./chunk-A3QHHUN5.js";

package/dist/worker_public/src/index.js

@@ -4,7 +4,7 @@ import {
   handleMemories,
   scoreFaithfulness,
   scoreJudge
-} from "../../chunk-GYCUMFOA.js";
+} from "../../chunk-FQVP3JUA.js";
 import {
   TOOLS_REGISTRY,
   bindModelNode,
@@ -12,7 +12,7 @@ import {
 } from "../../chunk-QNLFNEI4.js";
 import {
   handleSearch
-} from "../../chunk-OREMF5LK.js";
+} from "../../chunk-TFGJ6NYS.js";
 import {
   buildMessages,
   citations,
@@ -34,7 +34,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "../../chunk-BBKAETKM.js";
+} from "../../chunk-25ECO5M7.js";
 import {
   cfBlobs,
   embed,
@@ -996,7 +996,7 @@ async function callToolOnce(env, ctx, _req, tier, key, name, args, t0) {
       const r = await runTool({ ...env, DB: env.DB }, { name, args }, "mcp");
       payload = r ? { tool: r.name, query: r.query, result: r.output } : { error: { message: `tool ${name} returned nothing for the given arguments` } };
     } else {
-      const res = name === "ask" ? await (await import("../../ask-KP42T6DH.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-4CBKKNBJ.js")).handleSearch(env, ctx, inner, tier, key);
+      const res = name === "ask" ? await (await import("../../ask-E6YJUOIP.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-5SNP5CHL.js")).handleSearch(env, ctx, inner, tier, key);
       payload = await res.json().catch(() => ({ error: { message: "tool transport failed", status: res.status } }));
     }
     telemetry(env, ctx, tier, `mcp:${name}`, null, !payload?.error, JSON.stringify(payload).length, await sha256Hex(`${name}:${JSON.stringify(args ?? {})}`), void 0, void 0, { durationMs: Date.now() - t0 });

package/package.json

@@ -27,7 +27,7 @@
     "playwright": "^1.62.1",
     "yaml": "^2.9.1"
   },
-  "version": "0.2.100",
+  "version": "0.2.102",
   "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/lexical.ts

@@ -53,7 +53,7 @@ function rowMeta(r: any): ChunkMeta {
   };
 }
 
-function rowsToHits(rows: any[]): Hit[] {
+export function rowsToHits(rows: any[]): Hit[] {
   return rows.map((r: any) => {
     // rank is bm25 (lower better) → convert to positive score that RRF can ignore
     // (RRF uses rank position, not score). score kept for logging only.

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

@@ -54,10 +54,10 @@ export const conceptGraph: Stage = {
       // candidates rank by the corpus's own BM25 within the set — the
       // index filter this lane used returned empty for its whole life
       // (Vectorize metadata filtering is dead on this index)
-      const gc = await lexicalWithin(c.env, c.rq || c.query, [...numbers], 12);
+      const gc = await lexicalWithin(c.env, c.rq || c.query, [...numbers], 6);
       const seenIds0 = new Set(c.matches.map((m: any) => m.id));
       let merged0 = 0;
-      for (const m of gc.slice(0, 6)) {
+      for (const m of gc.slice(0, 3)) {
         if (!seenIds0.has(m.id)) {
           c.matches.push({ id: m.id, score: m.score * THRESHOLDS.conceptGraphDiscount, metadata: m.metadata });
           seenIds0.add(m.id);

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

@@ -4,18 +4,20 @@
 // saved), the doc/edition-filtered query with its guessed-pin drop and
 // sparse-filter widen, and the plain unfiltered query.
 import { LIMITS } from "../config.ts";
-import { toVectorizeFilter } from "../selfquery.ts";
 import type { Stage } from "./types.ts";
 
 export const dense: Stage = {
   name: "dense",
   run: async (c) => {
-    const { env, filter, filters, vector, opts, rq, folded } = c;
+    const { env, filters, vector, opts, rq, folded } = c;
+    // the index's metadata filtering is DEAD on this platform (measured
+    // 2026-10-05: filters return empty for values proven present), so
+    // the scope never rides the query — the matches filter IN CODE and
+    // the sparse widen still covers thin scopes
     const q: any = { topK: LIMITS.retrieveK, returnMetadata: "all" };
-    if (filter) q.filter = filter;
     const optimistic = opts.optimisticHits ?? [];
     const sameLane = rq === folded; // optimistic vector === this lane's vector
-    if (!filter && sameLane && optimistic.length) {
+    if (!filters?.doc_number && sameLane && optimistic.length) {
       // Option C fast path: the unfiltered dense query already ran
       // concurrently with understanding — same vector, same topK, no
       // filter. Reusing it skips one serial Vectorize round-trip with a
@@ -31,8 +33,13 @@ export const dense: Stage = {
       // REPLACE an identical query, never dilute a better one.
       return;
     }
-    if (filter) {
-      let matches: any[] = (await env.VECTORIZE.query(vector, q)).matches ?? [];
+    if (filters?.doc_number) {
+      const allMatches: any[] = (await env.VECTORIZE.query(vector, q)).matches ?? [];
+      let matches: any[] = allMatches;
+      const before = matches.length;
+      matches = matches.filter((m) => String(m.metadata?.doc_number ?? "") === filters.doc_number);
+      if (filters.edition) matches = matches.filter((m) => !m.metadata?.edition || String(m.metadata.edition) === filters.edition);
+      console.log("dense scope (in code):", before, "→", matches.length, "in doc", filters.doc_number, filters.edition ? `@${filters.edition}` : "");
       // an edition pin corroborated by (almost) nothing means the pin
       // was a guess — understanding emits editions for families it
       // mixes up (observed: R 76 pinned @2021, an R 60 year; the corpus
@@ -42,15 +49,13 @@ export const dense: Stage = {
       // the doc-only filter and let family-relative steering rank
       // editions downstream. A pin the user actually asked for
       // survives — its edition exists in the corpus.
+      // the same guess-drop, in code: a pinned edition that starved the
+      // scoped set drops to the doc-only scope
       if (filters && filters.edition && matches.length < 3) {
-        const docOnly = await env.VECTORIZE.query(vector, {
-          topK: LIMITS.retrieveK,
-          returnMetadata: "all",
-          filter: toVectorizeFilter({ doc_number: filters.doc_number }),
-        });
-        if ((docOnly.matches ?? []).length > matches.length) {
-          console.log("edition pin dropped:", filters.doc_number, "@", filters.edition, "→", docOnly.matches?.length ?? 0, "doc-scoped hits (edition not in corpus)");
-          matches = docOnly.matches ?? [];
+        const docOnly = allMatches.filter((m) => String(m.metadata?.doc_number ?? "") === filters.doc_number);
+        if (docOnly.length > matches.length) {
+          console.log("edition pin dropped:", filters.doc_number, "@", filters.edition, "→", docOnly.length, "doc-scoped hits (edition not in corpus)");
+          matches = docOnly;
           filters.edition = undefined;
         }
       }

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

@@ -7,7 +7,8 @@
 // left every 2021 chunk out). One registry read per query; at most
 // `maxDocs` documents covered; additive — any failure skips.
 import { THRESHOLDS } from "../config.ts";
-import { toHits, type Stage } from "./types.ts";
+import type { Stage } from "./types.ts";
+import { rowsToHits } from "../lexical.ts";
 
 const maxDocs = 2;
 
@@ -43,12 +44,17 @@ export const editionCover: Stage = {
     let added = 0;
     for (const w of want.slice(0, maxDocs)) {
       try {
-        const q = await c.env.VECTORIZE.query(c.vector, {
-          topK: 3,
-          returnMetadata: "all",
-          filter: { $and: [{ docidentifier: { $eq: w.di } }, { edition: { $eq: w.edition } }] },
-        });
-        const hits = toHits((q as any).matches ?? []).map((h) => ({ ...h, score: top * THRESHOLDS.editionCoverDiscount }));
+        // identity resolution, not index filtering: the index's metadata
+        // filters are dead on this platform (measured 2026-10-05 — the
+        // cover's $and docidentifier/edition query returned empty for
+        // its whole life), but the D1 chunks table IS the corpus — the
+        // current edition's chunks read directly, text and all
+        const rows = await c.env.DB.prepare(
+          "SELECT id, doc_id, docidentifier, doctype, doc_number, edition, language, clause_anchor, clause_title, status, superseded_by, corpus, tier, text FROM chunks WHERE docidentifier = ?1 AND edition = ?2 LIMIT 3",
+        )
+          .bind(w.di, w.edition)
+          .all();
+        const hits = rowsToHits(rows.results ?? []).map((h) => ({ ...h, score: top * THRESHOLDS.editionCoverDiscount }));
         c.hits.push(...hits);
         added += hits.length;
         console.log("edition cover:", w.di, w.edition, `+${hits.length}`);

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

@@ -23,9 +23,15 @@ export const graphLane: Stage = {
     const g = (await c.lane["graph-lane"]!) as any;
     const seenIds = new Set(c.matches.map((m: any) => m.id));
     let merged = 0;
-    for (const m of (g ?? []).slice(0, 10)) {
+    // narrow by the reverted experiment's own law (rag#137): candidate-
+    // family flooding dilutes the pool — five candidates, and at HALF
+    // the lane discount so only family text the cross-encoder genuinely
+    // prefers survives the cut (the first breathing gate measured the
+    // flood signature: faithfulness sagging across unrelated legs and
+    // the typed blocks displaced from the context)
+    for (const m of (g ?? []).slice(0, 5)) {
       if (!seenIds.has(m.id)) {
-        c.matches.push({ id: m.id, score: m.score * THRESHOLDS.graphLaneDiscount, metadata: m.metadata });
+        c.matches.push({ id: m.id, score: m.score * THRESHOLDS.graphLaneDiscount * 0.5, metadata: m.metadata });
         seenIds.add(m.id);
         merged++;
       }

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

@@ -18,6 +18,7 @@ import { glossary } from "./glossary.ts";
 import { conceptGraph } from "./conceptGraph.ts";
 import { graphLane } from "./graphLane.ts";
 import { licensedLane } from "./licensedLane.ts";
+import { licensedCover } from "./licensedCover.ts";
 import { multiQuery } from "./multiQuery.ts";
 import { subQuery } from "./subQuery.ts";
 import { poolOpen } from "./poolOpen.ts";
@@ -68,6 +69,7 @@ export const STAGES: Stage[] = [
   editionSteer,
   propagate,
   diversity,
+  licensedCover,
   typedPin,
   sectionDescent,
   dedup,

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

@@ -0,0 +1,31 @@
+// Licensed cover (the seat): an entitled, topically-matched caller
+// asked about the licensed standard, so the licensed package's own
+// machinery must be ABLE to reach the answer — the public restatements
+// crowd the window and the citation the golden legs assert rides this
+// passage. When the rerank cut drops every licensed unit, the lane's
+// best unit takes one seat at the end of the window (appended after
+// ranking, like the section-descent children — the window floor keeps
+// appended passages by construction). Narrow by law: one unit, only
+// when none is present, only for an armed licensed lane.
+import { portIndex } from "../env.ts";
+import { toHits } from "./types.ts";
+import type { Stage } from "./types.ts";
+
+export const licensedCover: Stage = {
+  name: "licensed-cover",
+  failure: "additive",
+  when: (c) => !!c.opts.licensedDocNumbers?.length && c.finalHits.length > 0,
+  run: async (c) => {
+    const keys = new Set(c.opts.licensedDocNumbers!);
+    const present = c.finalHits.some((h) => keys.has(String((h.metadata as any).standard ?? "")));
+    if (present) return;
+    const ids = ((await (c.lane["licensed-ids"] as Promise<string[]> | undefined)) ?? []).slice(0, 8);
+    if (!ids.length) return;
+    const matches = await portIndex(c.env, "public").getByIds(ids);
+    const unit = matches.find((m) => m.metadata && keys.has(String((m.metadata as any).standard ?? "")));
+    if (!unit) return;
+    const hit = toHits([unit as any])[0]!;
+    c.finalHits = [...c.finalHits, hit];
+    console.log("licensed cover: seated", unit.id, "—", (unit.metadata as any).docidentifier);
+  },
+};

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

@@ -44,13 +44,18 @@ export const licensedLane: Stage = {
     const matches = await portIndex(c.env, "public").getByIds(ids);
     const seenIds = new Set(c.matches.map((m: any) => m.id));
     let merged = 0;
+    const mergedIds: string[] = [];
     for (const m of matches) {
       if (m.metadata && !seenIds.has(m.id)) {
         c.matches.push({ id: m.id, score: THRESHOLDS.licensedLaneScore, metadata: m.metadata });
         seenIds.add(m.id);
         merged++;
+        mergedIds.push(m.id);
       }
     }
+    // the cover stage's stash: the licensed units' ids, so the seat can
+    // recover a unit the rerank cut dropped
+    c.lane["licensed-ids"] = Promise.resolve(mergedIds);
     console.log("licensed lane:", ids.length, "units,", matches.length, "vectors,", merged, "merged");
   },
 };