@konneal/engine 0.2.18 → 0.2.19

package/dist/{ask-4O753Y4T.js → ask-HCZ4CDUR.js}

@@ -1,6 +1,6 @@
 import {
   handleAsk
-} from "./chunk-PBXE2LFD.js";
+} from "./chunk-AN4VFBBZ.js";
 import "./chunk-7UT76RQO.js";
 import "./chunk-EFQALN2Z.js";
 import "./chunk-DBBGOOMZ.js";

package/dist/boundary.d.ts

@@ -0,0 +1,17 @@
+export interface LicensedEntry {
+    key: string;
+    package?: string;
+    doc_number?: string;
+    title?: string;
+}
+export interface LicensedMatch {
+    entry: LicensedEntry;
+    matched: string[];
+}
+export declare function distinctiveTokens(title: string): string[];
+/** The licensed entry whose distinctive title tokens best match the
+ *  question. Needs TWO token hits (or one hyphenated-compound hit) —
+ *  a single shared word is not a topic match. */
+export declare function matchLicensedTopic(query: string, licensed: LicensedEntry[]): LicensedMatch | null;
+/** The note text: the posture instruction for an unentitled match. */
+export declare function boundaryNoteText(match: LicensedMatch, citing: string[]): string;

package/dist/{chunk-PBXE2LFD.js → chunk-AN4VFBBZ.js}

@@ -1074,6 +1074,54 @@ function evaluateAggregation(nodes, query) {
   return null;
 }
 
+// workers/worker_public/src/boundary.ts
+var TITLE_STOPWORDS = /* @__PURE__ */ new Set([
+  "environmental",
+  "testing",
+  "test",
+  "tests",
+  "guidance",
+  "generic",
+  "standards",
+  "standard",
+  "electromagnetic",
+  "compatibility",
+  "environment",
+  "description",
+  "measurement",
+  "techniques",
+  "immunity",
+  "residential",
+  "commercial",
+  "industrial",
+  "environments",
+  "and",
+  "for",
+  "the",
+  "iec",
+  "iso"
+]);
+function distinctiveTokens(title) {
+  return title.toLowerCase().split(/[^a-z0-9.]+/).filter((w) => w.length >= 3 && !TITLE_STOPWORDS.has(w) && !/^[0-9:-]+$/.test(w));
+}
+function matchLicensedTopic(query, licensed) {
+  const words = new Set(query.toLowerCase().split(/[^a-z0-9.]+/).filter(Boolean));
+  let best = null;
+  for (const entry of licensed) {
+    if (!entry.title) continue;
+    const matched = distinctiveTokens(entry.title).filter((w) => words.has(w));
+    if (matched.length < 2) continue;
+    if (!best || matched.length > best.matched.length) best = { entry, matched };
+  }
+  return best;
+}
+function boundaryNoteText(match, citing) {
+  const doc = match.entry.doc_number ?? match.entry.key;
+  const title = match.entry.title ?? doc;
+  const refs = citing.length ? `The public corpus references it from ${citing.join(", ")}.` : "";
+  return `LICENSED BOUNDARY \u2014 ${title} (IEC ${doc}) is licensed content in this deployment; its procedure is NOT part of the public corpus you are grounded in. ${refs} When answering: name the licensed document as the authoritative source of the procedure and say it is available to entitled callers; do NOT recite its conditioning or severity parameters (specific temperatures, humidity levels, durations or cycle counts) as if from the source \u2014 describe only what the public grounding passages themselves state, attributed to their own publications.`;
+}
+
 // workers/worker_public/src/drafts.ts
 var ACT_VERB = "(?:draft|prepare|pre-?fill|fill\\s+(?:in|out)|start|submit|file|lodge)";
 var ACT_TARGET = "(?:new\\s+)?(?:certification\\s+|type[ -]evaluation\\s+|OIML[- ]CS\\s+)?application";
@@ -1963,6 +2011,24 @@ ${summary}` }] : [],
     }
   } : null;
   if (aggregationVerdict) console.log("aggregation engine:", aggregationVerdict.operation, aggregationVerdict.table, "\u2192", aggregationVerdict.value);
+  let boundaryNote = null;
+  if (P().sources?.licensed?.length) {
+    const match = matchLicensedTopic(q.query, P().sources.licensed);
+    if (match && !(standardKeys?.has(match.entry.key) ?? false)) {
+      const docNum = match.entry.doc_number ?? "";
+      let citing = [];
+      if (docNum) {
+        const rows = await env.DB.prepare(
+          "SELECT n.label AS label FROM graph_edges e JOIN graph_nodes n ON e.src = n.id WHERE e.kind = 'cites' AND e.dst LIKE ?1 LIMIT 4"
+        ).bind(`%${docNum}%`).all().catch(() => ({ results: [] }));
+        citing = (rows.results ?? []).map((r) => {
+          const m = String(r.label ?? "").match(/^OIML-([A-Z]+)-(\d+)(?:-([A-Za-z0-9]+))?-(\d{4})$/);
+          return m ? `OIML ${m[1]} ${m[2]}${m[3] ? `-${m[3]}` : ""} (${m[4]})` : String(r.label ?? "");
+        });
+      }
+      boundaryNote = boundaryNoteText(match, citing);
+    }
+  }
   try {
     const tR = Date.now();
     if (declaredCtx?.kind === "account") {
@@ -2056,7 +2122,7 @@ Answer account questions from these records ONLY: name the record when you use i
     q.lang,
     keptHistory,
     // stage-extracted graph facts (GraphRAG) ride the same note channel
-    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, licenseNote, ...retrieved.notes ?? []].filter(Boolean).join("\n") || void 0,
+    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...retrieved.notes ?? []].filter(Boolean).join("\n") || void 0,
     summary,
     budget
   );

package/dist/worker_public/src/index.js

@@ -7,7 +7,7 @@ import {
   handleMemories,
   scoreJudge,
   standardForDocNumber
-} from "../../chunk-PBXE2LFD.js";
+} from "../../chunk-AN4VFBBZ.js";
 import {
   buildMessages,
   citations,
@@ -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-4O753Y4T.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-RFKT7ZZY.js")).handleSearch(env, ctx, inner, tier, key);
+    const res = name === "ask" ? await (await import("../../ask-HCZ4CDUR.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-RFKT7ZZY.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

@@ -34,7 +34,7 @@
     "@oimlsmart/oiml-pubid": "^1.2.1"
   },
   "dependencies": {},
-  "version": "0.2.18",
+  "version": "0.2.19",
   "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

@@ -23,6 +23,7 @@ import { bindModelNode, licenseBoundaryNote, licenseBoundaryRefusal, licensedEnt
 import { evaluate as machineEvaluate, verdictNote } from "./verdict";
 import { evaluateConditionSets, quantitiesIn, type ConditionVerdict } from "./conditions";
 import { evaluateAggregation, type AggregationVerdict } from "./aggregation";
+import { matchLicensedTopic, boundaryNoteText } from "./boundary";
 import { detectDraftIntent, prepareDraft } from "./drafts";
 import { memoryNote } from "./memories";
 import { entitlementScope, resolveRequestScope, requestSalt } from "./requestScope";
@@ -846,6 +847,29 @@ async function handleAsk(
       }
     : null;
   if (aggregationVerdict) console.log("aggregation engine:", aggregationVerdict.operation, aggregationVerdict.table, "→", aggregationVerdict.value);
+  // ── the licensed boundary note (TODO.rag/12): an unentitled question
+  // that is topically ABOUT a licensed standard gets the boundary
+  // posture — the citation graph names the public publications that
+  // reference the licensed document, and the note instructs the model
+  // to attribute, never to recite the licensed parameters.
+  let boundaryNote: string | null = null;
+  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 ?? "";
+      let citing: string[] = [];
+      if (docNum) {
+        const rows = await env.DB.prepare(
+          "SELECT n.label AS label FROM graph_edges e JOIN graph_nodes n ON e.src = n.id WHERE e.kind = 'cites' AND e.dst LIKE ?1 LIMIT 4",
+        ).bind(`%${docNum}%`).all().catch(() => ({ results: [] }));
+        citing = (rows.results ?? []).map((r: any) => {
+          const m = String(r.label ?? "").match(/^OIML-([A-Z]+)-(\d+)(?:-([A-Za-z0-9]+))?-(\d{4})$/);
+          return m ? `OIML ${m[1]} ${m[2]}${m[3] ? `-${m[3]}` : ""} (${m[4]})` : String(r.label ?? "");
+        });
+      }
+      boundaryNote = boundaryNoteText(match, citing);
+    }
+  }
   try {
     const tR = Date.now();
     // ── The "my account" live read (TODO.ai-platform/03) — resolved
@@ -995,7 +1019,7 @@ async function handleAsk(
     q.lang,
     keptHistory,
     // stage-extracted graph facts (GraphRAG) ride the same note channel
-    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, licenseNote, ...(retrieved.notes ?? [])].filter(Boolean).join("\n") || undefined,
+    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...(retrieved.notes ?? [])].filter(Boolean).join("\n") || undefined,
     summary,
     budget,
   );

package/workers/worker_public/src/boundary.ts

@@ -0,0 +1,69 @@
+// The licensed boundary note (TODO.rag/12): when a question is about a
+// topic that matches a LICENSED standard the caller is not entitled to,
+// the note gives the answer model the boundary posture — name the
+// licensed document as the authoritative procedure, name the public
+// publications that reference it, and never recite the licensed
+// parameters as if from the source. Pure matching/text lives here; the
+// inverse-cites D1 read lives at the call site.
+
+export interface LicensedEntry {
+  key: string;
+  package?: string;
+  doc_number?: string;
+  title?: string;
+}
+
+export interface LicensedMatch {
+  entry: LicensedEntry;
+  matched: string[];
+}
+
+// title words that carry no topical identity — a question about "tests"
+// or "immunity" must not alone light the boundary
+const TITLE_STOPWORDS = new Set([
+  "environmental", "testing", "test", "tests", "guidance", "generic",
+  "standards", "standard", "electromagnetic", "compatibility",
+  "environment", "description", "measurement", "techniques", "immunity",
+  "residential", "commercial", "industrial", "environments", "and", "for", "the",
+  "iec", "iso",
+]);
+
+export function distinctiveTokens(title: string): string[] {
+  return title
+    .toLowerCase()
+    .split(/[^a-z0-9.]+/)
+    .filter((w) => w.length >= 3 && !TITLE_STOPWORDS.has(w) && !/^[0-9:-]+$/.test(w));
+}
+
+/** The licensed entry whose distinctive title tokens best match the
+ *  question. Needs TWO token hits (or one hyphenated-compound hit) —
+ *  a single shared word is not a topic match. */
+export function matchLicensedTopic(query: string, licensed: LicensedEntry[]): LicensedMatch | null {
+  const words = new Set(query.toLowerCase().split(/[^a-z0-9.]+/).filter(Boolean));
+  let best: { entry: LicensedEntry; matched: string[] } | null = null;
+  for (const entry of licensed) {
+    if (!entry.title) continue;
+    const matched = distinctiveTokens(entry.title).filter((w) => words.has(w));
+    if (matched.length < 2) continue;
+    if (!best || matched.length > best.matched.length) best = { entry, matched };
+  }
+  return best;
+}
+
+/** The note text: the posture instruction for an unentitled match. */
+export function boundaryNoteText(match: LicensedMatch, citing: string[]): string {
+  const doc = match.entry.doc_number ?? match.entry.key;
+  const title = match.entry.title ?? doc;
+  const refs = citing.length
+    ? `The public corpus references it from ${citing.join(", ")}.`
+    : "";
+  return (
+    `LICENSED BOUNDARY — ${title} (IEC ${doc}) is licensed content in this ` +
+    `deployment; its procedure is NOT part of the public corpus you are grounded in. ` +
+    `${refs} When answering: name the licensed document as the authoritative source of ` +
+    `the procedure and say it is available to entitled callers; do NOT recite its ` +
+    `conditioning or severity parameters (specific temperatures, humidity levels, ` +
+    `durations or cycle counts) as if from the source — describe only what the public ` +
+    `grounding passages themselves state, attributed to their own publications.`
+  );
+}