@konneal/engine 0.2.38 → 0.2.40

package/dist/{ask-VXV2BFNC.js → ask-QHODBUU2.js}

@@ -1,7 +1,7 @@
 import {
   handleAsk
-} from "./chunk-CNFQNCWA.js";
-import "./chunk-6GAYBMUM.js";
+} from "./chunk-5SLSRFOG.js";
+import "./chunk-N2UOENDY.js";
 import "./chunk-R2V3X6SQ.js";
 import "./chunk-A3QHHUN5.js";
 import "./chunk-4GJGBGJK.js";

package/dist/{chunk-CNFQNCWA.js → chunk-5SLSRFOG.js}

@@ -9,6 +9,7 @@ import {
   contextNote,
   editionNote,
   embed,
+  experimentalSourceLabels,
   generateOnce,
   graphExpand,
   identityNote,
@@ -34,7 +35,7 @@ import {
   syntheticUnderstanding,
   telemetry,
   understandQuery
-} from "./chunk-6GAYBMUM.js";
+} from "./chunk-N2UOENDY.js";
 import {
   corsHeaders,
   err,
@@ -2475,7 +2476,7 @@ Answer account questions from these records ONLY: name the record when you use i
 `));
           send({ type: "read", read: readAs() });
           const sourceQuality = answerQuality(cites.map((c) => c.quality));
-          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality) } : {}, ...liveRecords ? { records: liveRecords } : {}, quota });
+          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality), ...sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {} } : {}, ...liveRecords ? { records: liveRecords } : {}, quota });
           let full = "";
           try {
             for await (const tok of sseTokens(stream)) {
@@ -2609,7 +2610,7 @@ Answer account questions from these records ONLY: name the record when you use i
   }
   completionBlocks.push(...await completeFigures(env.DB, answer, [...c2ns.blocks, ...completionBlocks], used));
   const jsonQuality = answerQuality(finalCites.map((c) => c.quality));
-  const out = { answer, citations: finalCites, ...jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality) } : {}, model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...verdictBlock ? [verdictBlock] : [], ...conditionBlock ? [conditionBlock] : [], ...aggregationBlock ? [aggregationBlock] : [], ...completionBlocks], context_applied: ctxApplied, ...liveRecords ? { records: liveRecords } : {} };
+  const out = { answer, citations: finalCites, ...jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality), ...jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {} } : {}, model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...verdictBlock ? [verdictBlock] : [], ...conditionBlock ? [conditionBlock] : [], ...aggregationBlock ? [aggregationBlock] : [], ...completionBlocks], context_applied: ctxApplied, ...liveRecords ? { records: liveRecords } : {} };
   const cacheable = !contextual && !declaredCtx && !answer.includes(refusalAnswer()) && finalAnchors.violations.length === 0;
   if (cacheable) {
     const warmVec = await warmEmbed ?? null;

package/dist/{chunk-GDQOFP6B.js → chunk-AKAFLMIB.js}

@@ -7,7 +7,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "./chunk-6GAYBMUM.js";
+} from "./chunk-N2UOENDY.js";
 import {
   corsHeaders,
   err,

package/dist/{chunk-6GAYBMUM.js → chunk-N2UOENDY.js}

@@ -77,9 +77,21 @@ function answerQuality(qualities) {
   return worst;
 }
 function qualityNote(q) {
-  if (q === "verified") return "Grounded in verified sources: this publication's requirements ride a machine-checkable data model.";
-  if (q === "curated") return "Grounded in the edited corpus: Metanorma-authored documents, chunked at clause boundaries.";
-  return "Partly grounded in experimental OCR text: verify quotations and table values against the official publication.";
+  if (q === "verified") return "High confidence: this publication's requirements ride a machine-checkable data model.";
+  if (q === "curated") return "Established sources: Metanorma-edited documents, chunked at clause boundaries.";
+  return "WARNING: Partly grounded in experimental data source that was derived from OCR content. Please verify content against official publications.";
+}
+function experimentalSourceLabels(cites, cap = 6) {
+  const labels = [];
+  for (const c of cites) {
+    if (c.quality !== "ocr") continue;
+    const id = String(c.docidentifier || c.doc_id || "source");
+    const anchor = String(c.clause_anchor ?? "");
+    const garbage = /^[0-9a-f]{8}-[0-9a-f]{4}-/i.test(anchor) || anchor.startsWith("_") && anchor.length > 12;
+    labels.push(garbage || !anchor || anchor === "overview" ? id : `${id} \xA7${anchor}`);
+    if (labels.length >= cap) break;
+  }
+  return labels;
 }
 
 // workers/worker_public/src/lexical.ts
@@ -23836,13 +23848,23 @@ function opMemberCanWrite(member) {
   if (member.via !== "op-token") return true;
   return /\bwrite\b/.test(member.scope ?? "");
 }
-function opMemberFromIntrospection(answer, ids) {
+function opMemberFromIntrospection(answer, ids, patServices = []) {
   if (!answer?.active) return null;
-  const clientId = String(answer.client_id ?? "");
-  if (!ids.includes(clientId)) return null;
   const sub = String(answer.sub ?? "");
   if (!sub) return null;
-  return { sub, scope: String(answer.scope ?? ""), via: "op-token" };
+  const clientId = String(answer.client_id ?? "");
+  if (clientId) {
+    if (!ids.includes(clientId)) return null;
+    return { sub, scope: String(answer.scope ?? ""), via: "op-token" };
+  }
+  if (!patServices.length) return null;
+  const scope = String(answer.scope ?? "");
+  const services = scope.split(/\s+/).filter(Boolean).map((s8) => s8.split(":")[0]);
+  if (!services.some((s8) => patServices.includes(s8))) return null;
+  return { sub, scope, via: "op-token" };
+}
+function patServiceIds(env) {
+  return String(env.OIDC_PAT_SERVICES ?? "").split(/[\s,]+/).filter(Boolean);
 }
 async function opTokenMember(env, cfg, req) {
   const m = /^Bearer\s+(.+)$/i.exec((req.headers.get("Authorization") ?? "").trim());
@@ -23872,7 +23894,7 @@ async function opTokenMember(env, cfg, req) {
     } catch {
     }
   }
-  return opMemberFromIntrospection(answer, ids);
+  return opMemberFromIntrospection(answer, ids, patServiceIds(env));
 }
 
 // workers/worker_public/src/oidc.ts
@@ -24584,6 +24606,7 @@ export {
   generateOnce,
   answerQuality,
   qualityNote,
+  experimentalSourceLabels,
   ftsMatchQuery,
   refCodec,
   NO_CONTEXT,

package/dist/env.d.ts

@@ -36,6 +36,10 @@ export interface Env {
      *  deployment admits as member-read credentials via introspection.
      *  Empty/absent = the tier is off. */
     OIDC_DEVICE_CLIENT_IDS?: string;
+    /** the services whose OP-minted personal access tokens (the PAT form
+     *  of introspection — no client_id on the wire) this deployment admits
+     *  as member-read credentials. Empty/absent = PATs stay anonymous. */
+    OIDC_PAT_SERVICES?: string;
     SESSION_SECRET?: string;
     /** TODO.ai-platform/03 — the "my account" live-data delegation: the
      *  platform instance's API base + its client id at the OP (the

package/dist/livedata.d.ts

@@ -111,9 +111,17 @@ export declare function opMemberCanWrite(member: {
     scope?: string;
 } | null): boolean;
 /** The introspection answer → the member credential, judged against the
- *  deployment's allowlist. Inactive, foreign-client or subject-less
- *  answers all resolve to null — never to a widened guess. */
-export declare function opMemberFromIntrospection(answer: any, ids: string[]): OpMember | null;
+ *  deployment's two allowlists. The OP answers client-bearing tokens
+ *  WITH client_id (the access-token table, the machine JWTs) and its
+ *  PAT form WITHOUT one — a PAT's identity on the wire is its granted
+ *  SERVICES (the PAT grammar's whole vocabulary). So the deployment
+ *  names device-grant clients (OIDC_DEVICE_CLIENT_IDS) and, separately,
+ *  the services whose personal access tokens it admits
+ *  (OIDC_PAT_SERVICES). Both lists empty = the tier stays off; a token
+ *  matching neither list resolves to null — never a widened guess. */
+export declare function opMemberFromIntrospection(answer: any, ids: string[], patServices?: string[]): OpMember | null;
+/** The PAT-services allowlist (see opMemberFromIntrospection). */
+export declare function patServiceIds(env: any): string[];
 /** The Bearer read + introspection, KV-cached 45s keyed by the token's
  *  hash (an ask burst costs one introspection; a deactivation lands
  *  within a minute). The cached object is the OP's RAW answer — the

package/dist/quality.d.ts

@@ -2,5 +2,16 @@ import type { ChunkMeta } from "../../shared/chunk.ts";
 export type SourceQuality = "verified" | "curated" | "ocr";
 export declare function hitQuality(m: ChunkMeta): SourceQuality;
 export declare function answerQuality(qualities: (SourceQuality | undefined)[]): SourceQuality | null;
-/** The one-line posture the interface renders under the answer. */
+/** The one-line posture the interface renders under the answer. The
+ *  OCR rung is the owner's 2026-09-26 wording, verbatim. */
 export declare function qualityNote(q: SourceQuality): string;
+/** The citations standing on the OCR rung, named for the reader — the
+ *  confidence line says WHICH sources are experimental, not just that
+ *  some are. Producer-UUID anchors say nothing and are dropped; a cap
+ *  keeps the line readable on citation-heavy answers. */
+export declare function experimentalSourceLabels(cites: {
+    quality?: string;
+    docidentifier?: string;
+    doc_id?: string;
+    clause_anchor?: string;
+}[], cap?: number): string[];

package/dist/{search-HBE47CK6.js → search-QZDQVB57.js}

@@ -1,7 +1,7 @@
 import {
   handleSearch
-} from "./chunk-GDQOFP6B.js";
-import "./chunk-6GAYBMUM.js";
+} from "./chunk-AKAFLMIB.js";
+import "./chunk-N2UOENDY.js";
 import "./chunk-R2V3X6SQ.js";
 import "./chunk-A3QHHUN5.js";
 import "./chunk-4GJGBGJK.js";

package/dist/worker_public/src/index.js

@@ -1,6 +1,6 @@
 import {
   handleSearch
-} from "../../chunk-GDQOFP6B.js";
+} from "../../chunk-AKAFLMIB.js";
 import {
   bindModelNode,
   checkQuoteAnchors,
@@ -9,7 +9,7 @@ import {
   modelGroundingBlock,
   scoreJudge,
   standardForDocNumber
-} from "../../chunk-CNFQNCWA.js";
+} from "../../chunk-5SLSRFOG.js";
 import {
   buildMessages,
   citations,
@@ -35,7 +35,7 @@ import {
   sessionFrom,
   telemetry,
   understandQuery
-} from "../../chunk-6GAYBMUM.js";
+} from "../../chunk-N2UOENDY.js";
 import {
   authenticate,
   corsHeaders,
@@ -999,7 +999,7 @@ 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-VXV2BFNC.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-HBE47CK6.js")).handleSearch(env, ctx, inner, tier, key);
+    const res = name === "ask" ? await (await import("../../ask-QHODBUU2.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-QZDQVB57.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": "0.2.38",
+  "version": "0.2.40",
   "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

@@ -25,7 +25,7 @@ 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 { answerQuality, qualityNote } from "./quality";
+import { answerQuality, experimentalSourceLabels, qualityNote } from "./quality";
 import { detectDraftIntent, prepareDraft } from "./drafts";
 import { detectApiCallIntent, prepareApiCall } from "./apicalls";
 import { memoryNote } from "./memories";
@@ -1147,7 +1147,7 @@ async function handleAsk(
           // retrieval, before a single token of the answer
           send({ type: "read", read: readAs() });
           const sourceQuality = answerQuality(cites.map((c: any) => c.quality));
-          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...(sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality) } : {}), ...(liveRecords ? { records: liveRecords } : {}), quota, });
+          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...(sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality), ...(sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {}) } : {}), ...(liveRecords ? { records: liveRecords } : {}), quota, });
           let full = "";
           try {
             for await (const tok of sseTokens(stream)) {
@@ -1342,7 +1342,7 @@ async function handleAsk(
   completionBlocks.push(...(await completeFigures(env.DB, answer, [...c2ns.blocks, ...completionBlocks], used)));
 
   const jsonQuality = answerQuality(finalCites.map((c: any) => c.quality));
-  const out = { answer, citations: finalCites, ...(jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality) } : {}), model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...(verdictBlock ? [verdictBlock] : []), ...(conditionBlock ? [conditionBlock] : []), ...(aggregationBlock ? [aggregationBlock] : []), ...completionBlocks], context_applied: ctxApplied, ...(liveRecords ? { records: liveRecords } : {}) };
+  const out = { answer, citations: finalCites, ...(jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality), ...(jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {}) } : {}), model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...(verdictBlock ? [verdictBlock] : []), ...(conditionBlock ? [conditionBlock] : []), ...(aggregationBlock ? [aggregationBlock] : []), ...completionBlocks], context_applied: ctxApplied, ...(liveRecords ? { records: liveRecords } : {}) };
   const cacheable = !contextual && !declaredCtx && !answer.includes(refusalAnswer()) && finalAnchors.violations.length === 0;
   if (cacheable) {
     const warmVec = (await warmEmbed) ?? null;

package/workers/worker_public/src/env.ts

@@ -38,6 +38,10 @@ export interface Env {
    *  deployment admits as member-read credentials via introspection.
    *  Empty/absent = the tier is off. */
   OIDC_DEVICE_CLIENT_IDS?: string;
+  /** the services whose OP-minted personal access tokens (the PAT form
+   *  of introspection — no client_id on the wire) this deployment admits
+   *  as member-read credentials. Empty/absent = PATs stay anonymous. */
+  OIDC_PAT_SERVICES?: string;
   SESSION_SECRET?: string;
   /** TODO.ai-platform/03 — the "my account" live-data delegation: the
    *  platform instance's API base + its client id at the OP (the

package/workers/worker_public/src/livedata.ts

@@ -431,15 +431,35 @@ export function opMemberCanWrite(member: { via?: string; scope?: string } | null
 }
 
 /** The introspection answer → the member credential, judged against the
- *  deployment's allowlist. Inactive, foreign-client or subject-less
- *  answers all resolve to null — never to a widened guess. */
-export function opMemberFromIntrospection(answer: any, ids: string[]): OpMember | null {
+ *  deployment's two allowlists. The OP answers client-bearing tokens
+ *  WITH client_id (the access-token table, the machine JWTs) and its
+ *  PAT form WITHOUT one — a PAT's identity on the wire is its granted
+ *  SERVICES (the PAT grammar's whole vocabulary). So the deployment
+ *  names device-grant clients (OIDC_DEVICE_CLIENT_IDS) and, separately,
+ *  the services whose personal access tokens it admits
+ *  (OIDC_PAT_SERVICES). Both lists empty = the tier stays off; a token
+ *  matching neither list resolves to null — never a widened guess. */
+export function opMemberFromIntrospection(answer: any, ids: string[], patServices: string[] = []): OpMember | null {
   if (!answer?.active) return null;
-  const clientId = String(answer.client_id ?? "");
-  if (!ids.includes(clientId)) return null;
   const sub = String(answer.sub ?? "");
   if (!sub) return null;
-  return { sub, scope: String(answer.scope ?? ""), via: "op-token" };
+  const clientId = String(answer.client_id ?? "");
+  if (clientId) {
+    if (!ids.includes(clientId)) return null;
+    return { sub, scope: String(answer.scope ?? ""), via: "op-token" };
+  }
+  if (!patServices.length) return null;
+  const scope = String(answer.scope ?? "");
+  const services = scope.split(/\s+/).filter(Boolean).map((s) => s.split(":")[0]);
+  if (!services.some((s) => patServices.includes(s))) return null;
+  return { sub, scope, via: "op-token" };
+}
+
+/** The PAT-services allowlist (see opMemberFromIntrospection). */
+export function patServiceIds(env: any): string[] {
+  return String(env.OIDC_PAT_SERVICES ?? "")
+    .split(/[\s,]+/)
+    .filter(Boolean);
 }
 
 /** The Bearer read + introspection, KV-cached 45s keyed by the token's
@@ -482,5 +502,5 @@ export async function opTokenMember(
       /* cacheless = introspect per ask */
     }
   }
-  return opMemberFromIntrospection(answer, ids);
+  return opMemberFromIntrospection(answer, ids, patServiceIds(env));
 }

package/workers/worker_public/src/quality.ts

@@ -40,9 +40,30 @@ export function answerQuality(qualities: (SourceQuality | undefined)[]): SourceQ
   return worst;
 }
 
-/** The one-line posture the interface renders under the answer. */
+/** The one-line posture the interface renders under the answer. The
+ *  OCR rung is the owner's 2026-09-26 wording, verbatim. */
 export function qualityNote(q: SourceQuality): string {
-  if (q === "verified") return "Grounded in verified sources: this publication's requirements ride a machine-checkable data model.";
-  if (q === "curated") return "Grounded in the edited corpus: Metanorma-authored documents, chunked at clause boundaries.";
-  return "Partly grounded in experimental OCR text: verify quotations and table values against the official publication.";
+  if (q === "verified") return "High confidence: this publication's requirements ride a machine-checkable data model.";
+  if (q === "curated") return "Established sources: Metanorma-edited documents, chunked at clause boundaries.";
+  return "WARNING: Partly grounded in experimental data source that was derived from OCR content. Please verify content against official publications.";
+}
+
+/** The citations standing on the OCR rung, named for the reader — the
+ *  confidence line says WHICH sources are experimental, not just that
+ *  some are. Producer-UUID anchors say nothing and are dropped; a cap
+ *  keeps the line readable on citation-heavy answers. */
+export function experimentalSourceLabels(
+  cites: { quality?: string; docidentifier?: string; doc_id?: string; clause_anchor?: string }[],
+  cap = 6,
+): string[] {
+  const labels: string[] = [];
+  for (const c of cites) {
+    if (c.quality !== "ocr") continue;
+    const id = String(c.docidentifier || c.doc_id || "source");
+    const anchor = String(c.clause_anchor ?? "");
+    const garbage = /^[0-9a-f]{8}-[0-9a-f]{4}-/i.test(anchor) || (anchor.startsWith("_") && anchor.length > 12);
+    labels.push(garbage || !anchor || anchor === "overview" ? id : `${id} §${anchor}`);
+    if (labels.length >= cap) break;
+  }
+  return labels;
 }