@canonry/canonry 6.1.4 → 6.1.5

package/dist/{chunk-5LZLDXEN.js → chunk-4UEX44C3.js}

@@ -19,21 +19,21 @@ import {
   runWithUsageTags,
   saveConfigPatch,
   sentimentInstallReadiness
-} from "./chunk-YOMLHHJS.js";
+} from "./chunk-FRBKYQN7.js";
 import {
   computeFactorAverages,
   deleteSiteCrawlGraphLayout,
   persistSiteCrawlGraphLayout,
   prepareSiteCrawlGraphLayout,
   recordedSharePct
-} from "./chunk-HB6XZOZ6.js";
+} from "./chunk-MSFBNFTI.js";
 import {
   CanonryMcpToolNames,
   CliError,
   PACKAGE_VERSION,
   canonryMcpTools,
   isMachineFormat
-} from "./chunk-6CR7DWLU.js";
+} from "./chunk-5QH72CTN.js";
 import {
   CC_CACHE_DIR,
   DUCKDB_SPEC,
@@ -214,7 +214,7 @@ import {
   siteLivenessState,
   toAlertView,
   usageCounters
-} from "./chunk-N7MRAN3G.js";
+} from "./chunk-MTEVU6KG.js";
 import {
   AGENT_MEMORY_VALUE_MAX_BYTES,
   AGENT_NONE,
@@ -414,7 +414,7 @@ import {
   visibilityReportRequestSchema,
   winnabilityClassLabel,
   withRetry
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/runtime-context.ts
 function cliRuntimeContext(env = process.env, stdio = process) {
@@ -7830,7 +7830,7 @@ function nonBlank(value) {
   return trimmed ? trimmed : void 0;
 }
 function resolveBuildCommit(env = process.env) {
-  const embedded = true ? "5bfed51c11b95f292c522e93568efee2f164c4ed" : void 0;
+  const embedded = true ? "07710f04d6bdd8ec8b4374a93f0f3c3d1b56f498" : void 0;
   return nonBlank(embedded) ?? nonBlank(env.CANONRY_COMMIT);
 }
 function resolveInstanceIdentity(env = process.env) {
@@ -7845,6 +7845,7 @@ import crypto6 from "crypto";
 import fs4 from "fs";
 import path3 from "path";
 import os3 from "os";
+import { setImmediate as yieldToEventLoop } from "timers/promises";
 import { and as and3, asc as asc2, count, eq as eq3, inArray, isNotNull, isNull, ne, sql as sql2 } from "drizzle-orm";
 
 // src/run-telemetry.ts
@@ -8115,6 +8116,8 @@ async function runWithConcurrency(items, limit, worker) {
   await Promise.all(Array.from({ length: cap }, next));
 }
 var PROVIDER_FANOUT_DEFAULT = 8;
+var BATCH_INGEST_CHUNK_SIZE = 32;
+var BATCH_INGEST_PAYLOAD_BYTES = 1024 * 1024;
 var SYNC_SLOT_DISPATCH = {
   idempotent: false,
   mode: ProviderDispatchModes.sync,
@@ -9350,10 +9353,10 @@ var JobRunner = class {
   /**
    * Read an ended batch's results into its run: every line mapped by
    * `custom_id` to its slot, parsed by the adapter's own parser and recorded
-   * through `recordSlot`, exactly like a sync answer but at the batch tier.
+   * through the shared slot preparation, like a sync answer at the batch tier.
    *
-   * Idempotent across crashes. Each line's outcome is written as soon as it is
-   * handled and a handled line is skipped on the next pass; the insert does
+   * Idempotent across crashes. Bounded chunks commit each snapshot together
+   * with its outcome; handled lines are skipped on the next pass. The insert does
    * nothing on a slot that is already answered; the reservation of unbilled
    * lines is released under `quota_released`, in the same transaction that
    * marks the batch `ingested`. Answers are scored against the project's
@@ -9375,30 +9378,74 @@ var JobRunner = class {
     const requests = new Map(this.db.select().from(providerBatchRequests).where(eq3(providerBatchRequests.batchId, batch.id)).all().map((row) => [row.id, row]));
     const ctx = { ...this.buildRunRecordingContext(batch.runId, batch.projectId), onInserted: () => {
     } };
+    const pending = [];
+    let pendingBytes = 0;
+    const queued = /* @__PURE__ */ new Set();
+    const flush = () => {
+      if (pending.length === 0) return true;
+      const chunk = pending.splice(0);
+      pendingBytes = 0;
+      const settled = this.persistBatchLines(batch, chunk);
+      if (!settled) return false;
+      for (const result2 of settled) {
+        const request = requests.get(result2.line.requestId);
+        requests.set(request.id, { ...request, outcome: result2.outcome, error: result2.line.error });
+        result2.line.slot?.report(result2.inserted);
+        result2.line.report?.();
+      }
+      queued.clear();
+      return true;
+    };
     try {
+      let linesSinceYield = 0;
       for await (const line of capability.results(batch.providerBatchId, registered.config)) {
+        if (linesSinceYield === BATCH_INGEST_CHUNK_SIZE) {
+          await yieldToEventLoop();
+          linesSinceYield = 0;
+        }
+        linesSinceYield++;
         const request = requests.get(line.customId);
         if (!request) {
           log2.warn("batch.unknown-line", { batchId: batch.id, provider: batch.provider, customId: line.customId });
           continue;
         }
-        if (request.outcome !== null) continue;
-        if (this.db.select({ status: providerBatches.status }).from(providerBatches).where(eq3(providerBatches.id, batch.id)).get()?.status !== ProviderBatchStatuses.ended) {
-          return { kind: "cancelled" };
+        if (request.outcome !== null || queued.has(request.id)) continue;
+        if (this.db.select({ status: providerBatches.status }).from(providerBatches).where(eq3(providerBatches.id, batch.id)).get()?.status !== ProviderBatchStatuses.ended) return { kind: "cancelled" };
+        const prepared = await this.prepareBatchLine(ctx, registered, batch, request, line);
+        const payloadBytes = Buffer.byteLength(prepared.slot?.snapshot.rawResponse ?? "") + Buffer.byteLength(prepared.slot?.snapshot.answerText ?? "") + Buffer.byteLength(prepared.error ?? "");
+        if (pending.length > 0 && pendingBytes + payloadBytes > BATCH_INGEST_PAYLOAD_BYTES) {
+          if (!flush()) return { kind: "cancelled" };
+          await yieldToEventLoop();
+          linesSinceYield = 0;
+        }
+        pending.push(prepared);
+        pendingBytes += payloadBytes;
+        queued.add(request.id);
+        if (pending.length === BATCH_INGEST_CHUNK_SIZE || pendingBytes >= BATCH_INGEST_PAYLOAD_BYTES) {
+          if (!flush()) return { kind: "cancelled" };
+          await yieldToEventLoop();
+          linesSinceYield = 0;
         }
-        const settled = await this.ingestBatchLine(ctx, registered, batch, request, line);
-        this.db.update(providerBatchRequests).set({ outcome: settled.outcome, error: settled.error }).where(and3(eq3(providerBatchRequests.id, request.id), isNull(providerBatchRequests.outcome))).run();
-        requests.set(request.id, { ...request, outcome: settled.outcome });
       }
+      if (!flush()) return { kind: "cancelled" };
     } catch (err) {
-      if (!(err instanceof RunCancelledError)) throw err;
+      let failure2 = err;
+      if (!(failure2 instanceof RunCancelledError)) {
+        try {
+          if (!flush()) return { kind: "cancelled" };
+        } catch (flushError) {
+          failure2 = flushError;
+        }
+      }
+      if (!(failure2 instanceof RunCancelledError)) throw failure2;
       this.markBatchesCancelled([batch.id], "Cancelled because its run is no longer running.");
       return { kind: "cancelled" };
     }
     return this.completeBatchIngest(batch);
   }
-  async ingestBatchLine(ctx, registered, batch, request, line) {
-    if (line.type !== "succeeded") return { outcome: unansweredOutcome(line.type), error: line.error };
+  async prepareBatchLine(ctx, registered, batch, request, line) {
+    const base = { requestId: request.id, slot: null };
+    if (line.type !== "succeeded") return { ...base, outcome: unansweredOutcome(line.type), error: line.error };
     const { adapter } = registered;
     if (!adapter.parseTrackedQueryResponse) throw new Error(`Provider ${adapter.name} cannot read a batch answer`);
     let raw;
@@ -9406,34 +9453,67 @@ var JobRunner = class {
       raw = adapter.parseTrackedQueryResponse(line.body, resolveProviderModel(adapter.name, request.requestedModel));
     } catch (err) {
       const error = describeError(err);
-      log2.warn("batch.parse-failed", { runId: batch.runId, batchId: batch.id, provider: batch.provider, executionId: request.executionId, error });
-      return { outcome: ProviderBatchRequestOutcomes.parse_failed, error };
+      return {
+        ...base,
+        outcome: ProviderBatchRequestOutcomes.parse_failed,
+        error,
+        report: () => log2.warn("batch.parse-failed", { runId: batch.runId, batchId: batch.id, provider: batch.provider, executionId: request.executionId, error })
+      };
     }
-    const unit = {
+    const slot = await this.prepareSlot(ctx, registered, {
       executionId: request.executionId,
       queryText: request.queryText,
       context: request.requestedContext,
       queryId: request.queryId,
       requestedModel: request.requestedModel
-    };
-    const written = await this.recordSlot(ctx, registered, unit, raw, {
+    }, raw, {
       idempotent: true,
       mode: ProviderDispatchModes.batch,
       providerBatchId: batch.id,
       pricingTier: PricingTiers.batch
     });
-    if (written) return { outcome: ProviderBatchRequestOutcomes.recorded, error: null };
-    const existing = this.db.select({ providerBatchId: querySnapshots.providerBatchId }).from(querySnapshots).where(and3(
-      eq3(querySnapshots.runId, batch.runId),
-      eq3(querySnapshots.measurementExecutionId, request.executionId),
-      eq3(querySnapshots.provider, registered.adapter.name)
-    )).get();
-    return existing?.providerBatchId === batch.id ? { outcome: ProviderBatchRequestOutcomes.recorded, error: null } : { outcome: ProviderBatchRequestOutcomes.duplicate, error: null };
+    return { requestId: request.id, slot, outcome: ProviderBatchRequestOutcomes.recorded, error: null };
+  }
+  /** Snapshots and their replay ledger commit together, with no callbacks or awaits. */
+  persistBatchLines(batch, lines) {
+    return this.db.transaction((tx) => {
+      if (tx.select({ status: providerBatches.status }).from(providerBatches).where(eq3(providerBatches.id, batch.id)).get()?.status !== ProviderBatchStatuses.ended) return null;
+      if (tx.select({ status: runs.status }).from(runs).where(eq3(runs.id, batch.runId)).get()?.status !== RunStatuses.running) throw new RunCancelledError(batch.runId);
+      const current = new Map(tx.select({
+        id: providerBatchRequests.id,
+        outcome: providerBatchRequests.outcome,
+        queryId: providerBatchRequests.queryId,
+        executionId: providerBatchRequests.executionId
+      }).from(providerBatchRequests).where(inArray(providerBatchRequests.id, lines.map((line) => line.requestId))).all().map((row) => [row.id, row]));
+      const settled = [];
+      for (const line of lines) {
+        const request = current.get(line.requestId);
+        if (!request || request.outcome !== null) continue;
+        current.delete(line.requestId);
+        let outcome = line.outcome;
+        let inserted = false;
+        if (line.slot) {
+          const snapshot = { ...line.slot.snapshot, queryId: request.queryId };
+          inserted = tx.insert(querySnapshots).values(snapshot).onConflictDoNothing().run().changes > 0;
+          if (!inserted) {
+            const existing = tx.select({ providerBatchId: querySnapshots.providerBatchId }).from(querySnapshots).where(and3(
+              eq3(querySnapshots.runId, batch.runId),
+              eq3(querySnapshots.measurementExecutionId, request.executionId),
+              eq3(querySnapshots.provider, batch.provider)
+            )).get();
+            outcome = existing?.providerBatchId === batch.id ? ProviderBatchRequestOutcomes.recorded : ProviderBatchRequestOutcomes.duplicate;
+          }
+        }
+        tx.update(providerBatchRequests).set({ outcome, error: line.error }).where(and3(eq3(providerBatchRequests.id, request.id), isNull(providerBatchRequests.outcome))).run();
+        settled.push({ line, outcome, inserted });
+      }
+      return settled;
+    }, { behavior: "immediate" });
   }
   /** Close an ingest: counts, the unbilled lines' quota, and `ingested`, in one transaction. */
   completeBatchIngest(batch) {
     const at = (/* @__PURE__ */ new Date()).toISOString();
-    return this.db.transaction((tx) => {
+    const result2 = this.db.transaction((tx) => {
       const txDb = tx;
       const current = txDb.select().from(providerBatches).where(eq3(providerBatches.id, batch.id)).get();
       if (current?.status !== ProviderBatchStatuses.ended) return { kind: "skipped" };
@@ -9449,9 +9529,19 @@ var JobRunner = class {
         updatedAt: at
       }).where(and3(eq3(providerBatches.id, batch.id), eq3(providerBatches.status, ProviderBatchStatuses.ended))).run();
       const { recordedCount: recorded, notRecorded } = tally;
-      log2.info("batch.ingested", { runId: batch.runId, batchId: batch.id, provider: batch.provider, recorded, notRecorded, released });
       return { kind: "ingested", recorded, notRecorded, released };
     });
+    if (result2.kind === "ingested") {
+      log2.info("batch.ingested", {
+        runId: batch.runId,
+        batchId: batch.id,
+        provider: batch.provider,
+        recorded: result2.recorded,
+        notRecorded: result2.notRecorded,
+        released: result2.released
+      });
+    }
+    return result2;
   }
   /**
    * Give up on an ended batch whose results cannot be read in time. It
@@ -9742,6 +9832,15 @@ var JobRunner = class {
    * writes nothing and resolves false.
    */
   async recordSlot(ctx, registeredProvider, unit, raw, dispatch) {
+    const prepared = await this.prepareSlot(ctx, registeredProvider, unit, raw, dispatch);
+    this.throwIfRunCancelled(ctx.runId);
+    const insert = this.db.insert(querySnapshots).values(prepared.snapshot);
+    const written = ctx.fill || dispatch.idempotent ? insert.onConflictDoNothing().run() : insert.run();
+    prepared.report(written.changes > 0);
+    return written.changes > 0;
+  }
+  /** Resolve network/file evidence and derived fields before opening a write transaction. */
+  async prepareSlot(ctx, registeredProvider, unit, raw, dispatch) {
     const { adapter } = registeredProvider;
     const providerName = adapter.name;
     const { domains: competitorDomains, aliases: competitorAliases } = ctx.competitorsFor(unit.executionId);
@@ -9798,7 +9897,7 @@ var JobRunner = class {
       fs4.renameSync(raw.screenshotPath, destPath);
       screenshotRelPath = `${ctx.runId}/${snapshotId}.png`;
     }
-    const insert = this.db.insert(querySnapshots).values({
+    const snapshot = {
       id: snapshotId,
       runId: ctx.runId,
       queryId: unit.queryId,
@@ -9850,20 +9949,23 @@ var JobRunner = class {
         overrides: registeredProvider.config.pricing
       }),
       createdAt: (/* @__PURE__ */ new Date()).toISOString()
-    });
-    const written = ctx.fill || dispatch.idempotent ? insert.onConflictDoNothing().run() : insert.run();
-    if (written.changes > 0) ctx.onInserted();
-    ctx.fill?.onOutcome(providerName, true);
-    log2.info("query.citation", {
-      runId: ctx.runId,
-      provider: providerName,
-      query: unit.queryText,
-      executionId: unit.executionId,
-      location: requestedContext?.label ?? null,
-      citationState,
-      answerMentioned
-    });
-    return written.changes > 0;
+    };
+    return {
+      snapshot,
+      report: (inserted) => {
+        if (inserted) ctx.onInserted();
+        ctx.fill?.onOutcome(providerName, true);
+        log2.info("query.citation", {
+          runId: ctx.runId,
+          provider: providerName,
+          query: unit.queryText,
+          executionId: unit.executionId,
+          location: requestedContext?.label ?? null,
+          citationState,
+          answerMentioned
+        });
+      }
+    };
   }
   /**
    * Rebuild a run's recording identity from the database, for a path that
@@ -13968,8 +14070,10 @@ async function executeSiteAudit(db, runId, projectId, opts = {}) {
 }
 
 // src/commands/backfill.ts
-import { and as and15, eq as eq17, inArray as inArray7, isNull as isNull2, sql as sql7 } from "drizzle-orm";
+import { setImmediate as yieldToEventLoop2 } from "timers/promises";
+import { and as and15, asc as asc4, desc as desc7, eq as eq17, gt as gt3, inArray as inArray7, isNull as isNull2, lte as lte2, sql as sql7 } from "drizzle-orm";
 var SNAPSHOT_BATCH_SIZE = 500;
+var PATH_REPAIR_PAGE_SIZE = 128;
 async function backfillAnswerVisibilityCommand(opts) {
   const config = loadConfig();
   const db = createClient(config.database);
@@ -14150,35 +14254,54 @@ No DB writes performed. Re-run without --dry-run to apply.`);
   }
 }
 function backfillNormalizedPaths(db, opts) {
-  const baseConditions = [];
-  if (opts?.projectId) {
-    baseConditions.push(eq17(gaTrafficSnapshots.projectId, opts.projectId));
-  }
-  const rows = db.select({
-    id: gaTrafficSnapshots.id,
-    landingPage: gaTrafficSnapshots.landingPage,
-    landingPageNormalized: gaTrafficSnapshots.landingPageNormalized
-  }).from(gaTrafficSnapshots).where(baseConditions.length > 0 ? and15(...baseConditions) : void 0).all();
+  return repairStoredPaths(db, gaTrafficSnapshots, opts);
+}
+async function repairStoredPaths(db, table, opts) {
+  const pageSize = opts?.pageSize ?? PATH_REPAIR_PAGE_SIZE;
+  const projectScope = opts?.projectId ? sql7`+${table.projectId} = ${opts.projectId}` : void 0;
+  const lastId = db.select({ id: table.id }).from(table).where(projectScope).orderBy(desc7(table.id)).limit(1).get()?.id;
+  if (lastId === void 0) return { examined: 0, updated: 0, unchanged: 0 };
+  let cursor;
+  let examined = 0;
   let updated = 0;
-  let unchanged = 0;
-  if (rows.length > 0) {
-    db.transaction((tx) => {
-      for (const row of rows) {
-        const next = normalizeUrlPath(row.landingPage);
-        if (next === null) {
-          unchanged++;
-          continue;
-        }
-        if (row.landingPageNormalized === next) {
-          unchanged++;
-          continue;
-        }
-        tx.update(gaTrafficSnapshots).set({ landingPageNormalized: next }).where(eq17(gaTrafficSnapshots.id, row.id)).run();
-        updated++;
-      }
+  for (; ; ) {
+    const rows = db.select({
+      id: table.id,
+      projectId: table.projectId,
+      landingPage: table.landingPage,
+      landingPageNormalized: table.landingPageNormalized
+    }).from(table).where(and15(
+      projectScope,
+      lte2(table.id, lastId),
+      cursor === void 0 ? void 0 : gt3(table.id, cursor)
+    )).orderBy(asc4(table.id)).limit(pageSize).all();
+    examined += rows.length;
+    const changes = rows.flatMap((row) => {
+      const next = normalizeUrlPath(row.landingPage);
+      return next === null || row.landingPageNormalized === next ? [] : [{ row, next }];
     });
+    if (changes.length > 0) {
+      const written = db.transaction((tx) => {
+        let written2 = 0;
+        for (const { row, next } of changes) {
+          written2 += tx.update(table).set({ landingPageNormalized: next }).where(and15(
+            eq17(table.id, row.id),
+            eq17(table.projectId, row.projectId),
+            eq17(table.landingPage, row.landingPage),
+            row.landingPageNormalized === null ? isNull2(table.landingPageNormalized) : eq17(table.landingPageNormalized, row.landingPageNormalized)
+          )).run().changes;
+        }
+        return written2;
+      });
+      updated += written;
+      if (written < changes.length) opts?.onConflict?.();
+    }
+    if (rows.length < pageSize) break;
+    cursor = rows.at(-1).id;
+    if (cursor === lastId) break;
+    await yieldToEventLoop2();
   }
-  return { examined: rows.length, updated, unchanged };
+  return { examined, updated, unchanged: examined - updated };
 }
 async function backfillNormalizedPathsCommand(opts) {
   const config = loadConfig();
@@ -14204,7 +14327,7 @@ async function backfillNormalizedPathsCommand(opts) {
     }
     projectId = project.id;
   }
-  const { examined, updated, unchanged } = backfillNormalizedPaths(db, { projectId });
+  const { examined, updated, unchanged } = await backfillNormalizedPaths(db, { projectId });
   const result2 = {
     project: projectFilter ?? null,
     examined,
@@ -14222,35 +14345,7 @@ async function backfillNormalizedPathsCommand(opts) {
   console.log(`  Unchanged: ${unchanged}`);
 }
 function backfillAiReferralPaths(db, opts) {
-  const baseConditions = [];
-  if (opts?.projectId) {
-    baseConditions.push(eq17(gaAiReferrals.projectId, opts.projectId));
-  }
-  const rows = db.select({
-    id: gaAiReferrals.id,
-    landingPage: gaAiReferrals.landingPage,
-    landingPageNormalized: gaAiReferrals.landingPageNormalized
-  }).from(gaAiReferrals).where(baseConditions.length > 0 ? and15(...baseConditions) : void 0).all();
-  let updated = 0;
-  let unchanged = 0;
-  if (rows.length > 0) {
-    db.transaction((tx) => {
-      for (const row of rows) {
-        const next = normalizeUrlPath(row.landingPage);
-        if (next === null) {
-          unchanged++;
-          continue;
-        }
-        if (row.landingPageNormalized === next) {
-          unchanged++;
-          continue;
-        }
-        tx.update(gaAiReferrals).set({ landingPageNormalized: next }).where(eq17(gaAiReferrals.id, row.id)).run();
-        updated++;
-      }
-    });
-  }
-  return { examined: rows.length, updated, unchanged };
+  return repairStoredPaths(db, gaAiReferrals, opts);
 }
 async function backfillAiReferralPathsCommand(opts) {
   const config = loadConfig();
@@ -14276,7 +14371,7 @@ async function backfillAiReferralPathsCommand(opts) {
     }
     projectId = project.id;
   }
-  const { examined, updated, unchanged } = backfillAiReferralPaths(db, { projectId });
+  const { examined, updated, unchanged } = await backfillAiReferralPaths(db, { projectId });
   const result2 = {
     project: projectFilter ?? null,
     examined,
@@ -14454,7 +14549,7 @@ function readStoredGroundingSources(rawResponse) {
   return result2;
 }
 async function backfillInsightsCommand(project, opts) {
-  const { IntelligenceService: IntelligenceService2 } = await import("./intelligence-service-QUAOWEIA.js");
+  const { IntelligenceService: IntelligenceService2 } = await import("./intelligence-service-JYGP6LME.js");
   const config = loadConfig();
   const db = createClient(config.database);
   migrate(db);
@@ -15871,7 +15966,7 @@ async function refreshAllIntegrations(client, projectName, now = /* @__PURE__ */
 }
 
 // src/notifier.ts
-import { eq as eq20, desc as desc7, and as and18, inArray as inArray10, or } from "drizzle-orm";
+import { eq as eq20, desc as desc8, and as and18, inArray as inArray10, or } from "drizzle-orm";
 import crypto21 from "crypto";
 var log19 = createLogger("Notifier");
 var MATERIAL_MAGNITUDE_DELTA = 20;
@@ -16412,7 +16507,7 @@ var Notifier = class {
         eq20(runs.kind, thisRun.kind),
         or(eq20(runs.status, "completed"), eq20(runs.status, "partial"))
       )
-    ).orderBy(desc7(runs.createdAt), desc7(runs.id)).limit(RECENT_FETCH_LIMIT).all();
+    ).orderBy(desc8(runs.createdAt), desc8(runs.id)).limit(RECENT_FETCH_LIMIT).all();
     const groups = groupRunsByCreatedAt(recentRuns);
     const currentGroupIdx = groups.findIndex((g) => g[0]?.createdAt === thisRun.createdAt);
     if (currentGroupIdx < 0) return [];
@@ -18631,7 +18726,7 @@ async function loadExternalMcpTools(servers, opts = {}) {
 
 // src/agent/memory-store.ts
 import crypto24 from "crypto";
-import { and as and19, desc as desc8, eq as eq22, like, sql as sql10, or as or2 } from "drizzle-orm";
+import { and as and19, desc as desc9, eq as eq22, like, sql as sql10, or as or2 } from "drizzle-orm";
 var COMPACTION_KEY_PREFIX = "compaction:";
 var COMPACTION_NOTES_PER_SESSION = 3;
 function rowToDto(row) {
@@ -18645,7 +18740,7 @@ function rowToDto(row) {
   };
 }
 function listMemoryEntries(db, projectId, opts = {}) {
-  const query = db.select().from(agentMemory).where(eq22(agentMemory.projectId, projectId)).orderBy(desc8(agentMemory.updatedAt));
+  const query = db.select().from(agentMemory).where(eq22(agentMemory.projectId, projectId)).orderBy(desc9(agentMemory.updatedAt));
   const rows = opts.limit === void 0 ? query.all() : query.limit(opts.limit).all();
   return rows.map(rowToDto);
 }
@@ -18691,7 +18786,7 @@ function loadRecentForHydrate(db, projectId, limit, sessionId) {
   return db.select().from(agentMemory).where(and19(
     eq22(agentMemory.projectId, projectId),
     or2(sql10`substr(${agentMemory.key}, 1, ${COMPACTION_KEY_PREFIX.length}) <> ${COMPACTION_KEY_PREFIX}`, sessionId ? sql10`substr(${agentMemory.key}, 1, ${sessionPrefix.length}) = ${sessionPrefix}` : void 0)
-  )).orderBy(desc8(agentMemory.updatedAt)).limit(limit).all().map(rowToDto);
+  )).orderBy(desc9(agentMemory.updatedAt)).limit(limit).all().map(rowToDto);
 }
 function writeCompactionNote(db, args) {
   if (Buffer.byteLength(args.summary, "utf8") > AGENT_MEMORY_VALUE_MAX_BYTES) {
@@ -18719,7 +18814,7 @@ function writeCompactionNote(db, args) {
         eq22(agentMemory.projectId, args.projectId),
         like(agentMemory.key, `${sessionPrefix}%`)
       )
-    ).orderBy(desc8(agentMemory.updatedAt)).all();
+    ).orderBy(desc9(agentMemory.updatedAt)).all();
     const stale = existing.slice(COMPACTION_NOTES_PER_SESSION).map((r) => r.id);
     if (stale.length > 0) {
       tx.delete(agentMemory).where(sql10`${agentMemory.id} IN (${sql10.join(stale.map((s) => sql10`${s}`), sql10`, `)})`).run();

package/dist/{chunk-6CR7DWLU.js → chunk-5QH72CTN.js}

@@ -128,7 +128,7 @@ import {
   trafficSeriesGranularitySchema,
   visibilityCompareSelectionSchema,
   visibilityReportRequestSchema
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/cli-error.ts
 function isMachineFormat(format) {

package/dist/{chunk-YOMLHHJS.js → chunk-FRBKYQN7.js}

@@ -5,7 +5,7 @@ import {
   PACKAGE_VERSION,
   canonryMcpTools,
   isMachineFormat
-} from "./chunk-6CR7DWLU.js";
+} from "./chunk-5QH72CTN.js";
 import {
   AGENT_PROVIDER_IDS,
   CANONRY_NPM_PACKAGE_URL,
@@ -50,7 +50,7 @@ import {
   skillsClientSchema,
   upgradeCaveatFor,
   upgradeCommandFor
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/config.ts
 import fs from "fs";

package/dist/{chunk-4MWLHTRS.js → chunk-J7LV4CS2.js}

@@ -3481,6 +3481,7 @@ function extractDomainsFromText(text) {
 function textContainsDomain(text, domain) {
   return extractDomainsFromText(text).some((candidate) => hostMatchesDomain(candidate, domain));
 }
+var URL_PATH_NORMALIZATION_VERSION = 1;
 function normalizeUrlPath(input) {
   if (input == null) return null;
   let trimmed = input.trim();
@@ -21378,6 +21379,7 @@ export {
   hostMatchesAnyDomain,
   extractDomainsFromText,
   textContainsDomain,
+  URL_PATH_NORMALIZATION_VERSION,
   normalizeUrlPath,
   brandKeyFromText,
   prepareBrandMatchText,

package/dist/{chunk-HB6XZOZ6.js → chunk-MSFBNFTI.js}

@@ -5,7 +5,7 @@ import {
   siteCrawlGraphNodes,
   siteCrawlPages,
   siteCrawlSnapshots
-} from "./chunk-N7MRAN3G.js";
+} from "./chunk-MTEVU6KG.js";
 import {
   SITE_CRAWL_GRAPH_MAX_EDGES,
   SITE_CRAWL_GRAPH_MAX_NODES,
@@ -13,7 +13,7 @@ import {
   TEMPLATE_LINK_RATIO_THRESHOLD,
   factorStatusFromScore,
   roundPreservingTotal
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/site-audit-factors.ts
 function recordedSharePct(factor) {

package/dist/{chunk-N7MRAN3G.js → chunk-MTEVU6KG.js}

@@ -869,7 +869,7 @@ import {
   wordpressSchemaDeployResultDtoSchema,
   wordpressSchemaStatusResultDtoSchema,
   wordpressStatusDtoSchema
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/intelligence-service.ts
 import { eq as eq82, desc as desc36, asc as asc15, and as and69, ne as ne14, or as or20, inArray as inArray32, gte as gte22, lte as lte20, isNull as isNull16, sql as sql31, exists } from "drizzle-orm";
@@ -912,6 +912,7 @@ __export(schema_exports, {
   conversionTrackingContracts: () => conversionTrackingContracts,
   crawlerEventsHourly: () => crawlerEventsHourly,
   crawlerVerificationManifestsHourly: () => crawlerVerificationManifestsHourly,
+  dataRepairCompletions: () => dataRepairCompletions,
   discoveryProbes: () => discoveryProbes,
   discoverySessions: () => discoverySessions,
   doctorHealthState: () => doctorHealthState,
@@ -1527,6 +1528,11 @@ var runtimeLogMetadata = sqliteTable("runtime_log_metadata", {
   dropped: integer("dropped").notNull().default(0),
   captureErrors: integer("capture_errors").notNull().default(0)
 });
+var dataRepairCompletions = sqliteTable("data_repair_completions", {
+  name: text("name").primaryKey().notNull(),
+  version: integer("version").notNull(),
+  completedAt: text("completed_at").notNull()
+});
 var apiKeys = sqliteTable("api_keys", {
   id: text("id").primaryKey(),
   name: text("name").notNull(),
@@ -8191,6 +8197,17 @@ var MIGRATION_VERSIONS = [
     statements: [
       `ALTER TABLE projects ADD COLUMN qualified_aliases TEXT NOT NULL DEFAULT '[]'`
     ]
+  },
+  {
+    version: 167,
+    name: "data-repair-completions",
+    statements: [
+      `CREATE TABLE IF NOT EXISTS data_repair_completions (
+        name TEXT PRIMARY KEY NOT NULL,
+        version INTEGER NOT NULL,
+        completed_at TEXT NOT NULL
+      )`
+    ]
   }
 ];
 var AERO_LEDGER_REBUILDS = [
@@ -8578,11 +8595,11 @@ var OperationalLogStore = class {
   append(event) {
     try {
       const entry = projectEvent(event);
-      if (Buffer.byteLength(JSON.stringify(entry), "utf8") > MAX_ENTRY_BYTES) {
+      const entryBytes = Buffer.byteLength(JSON.stringify(entry), "utf8");
+      if (entryBytes > MAX_ENTRY_BYTES) {
         throw new RangeError(`Operational log entry exceeds ${MAX_ENTRY_BYTES} bytes.`);
       }
       this.withoutLockWait(() => this.db.transaction((tx) => {
-        this.prune(tx);
         const metadata2 = readMetadata(tx);
         tx.run(sql3`
           INSERT INTO runtime_logs (
@@ -8590,12 +8607,12 @@ var OperationalLogStore = class {
           ) VALUES (
             ${metadata2.nextSequence}, ${entry.ts}, ${entry.level}, ${entry.module}, ${entry.action}, ${entry.message ?? null},
             ${entry.projectId ?? null}, ${entry.runId ?? null}, ${entry.context.actor ?? null}, ${entry.context.requestId ?? null},
-            ${JSON.stringify(entry.context)}, ${Buffer.byteLength(JSON.stringify(entry), "utf8")}
+            ${JSON.stringify(entry.context)}, ${entryBytes}
           )
         `);
         tx.run(sql3`UPDATE runtime_log_metadata SET next_sequence = ${metadata2.nextSequence + 1} WHERE id = ${METADATA_ID}`);
         this.prune(tx);
-      }));
+      }, { behavior: "immediate" }));
     } catch {
       this.recordCaptureError();
     }
@@ -92232,6 +92249,7 @@ export {
   providerBatches,
   providerBatchRequests,
   auditLog,
+  dataRepairCompletions,
   apiKeys,
   schedules,
   insightNotifyState,

package/dist/{chunk-TBAVVIVI.js → chunk-PAYC6TQ2.js}

@@ -4,14 +4,14 @@ import {
   installSkills,
   loadConfigRaw,
   saveConfigPatch
-} from "./chunk-YOMLHHJS.js";
+} from "./chunk-FRBKYQN7.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-6CR7DWLU.js";
+} from "./chunk-5QH72CTN.js";
 import {
   SKILL_MANIFEST_FILENAME,
   SkillsClients
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/skills-autosync.ts
 import fs from "fs";

package/dist/cli.js

@@ -35,11 +35,11 @@ import {
   trackCliCommandFinished,
   trackEvent,
   waitForServerRuntimeStartup
-} from "./chunk-5LZLDXEN.js";
+} from "./chunk-4UEX44C3.js";
 import {
   autoSyncSkills,
   formatAutoSyncNotice
-} from "./chunk-TBAVVIVI.js";
+} from "./chunk-PAYC6TQ2.js";
 import {
   configExists,
   createApiClient,
@@ -56,13 +56,13 @@ import {
   saveConfig,
   saveConfigPatch,
   shouldPrintServeSkillsNudge
-} from "./chunk-YOMLHHJS.js";
+} from "./chunk-FRBKYQN7.js";
 import {
   DEFAULT_DEMO_TRUSTED_PROXIES,
   trustedProxiesProblem,
   trustedProxyProblem
 } from "./chunk-RB2F4D3Z.js";
-import "./chunk-HB6XZOZ6.js";
+import "./chunk-MSFBNFTI.js";
 import {
   CliError,
   EXIT_SYSTEM_ERROR,
@@ -74,7 +74,7 @@ import {
   runMeasurementDraftAction,
   systemError,
   usageError
-} from "./chunk-6CR7DWLU.js";
+} from "./chunk-5QH72CTN.js";
 import {
   CLOUDFLARE_WORKER_BINDINGS,
   CLOUDFLARE_WORKER_GENERATED_MARKER,
@@ -82,6 +82,7 @@ import {
   DEFAULT_BOT_LIST,
   apiKeys,
   createClient,
+  dataRepairCompletions,
   generateWorkerScript,
   generateWranglerToml,
   migrate,
@@ -89,7 +90,7 @@ import {
   queries,
   renderReportHtml,
   runs
-} from "./chunk-N7MRAN3G.js";
+} from "./chunk-MTEVU6KG.js";
 import {
   AdsDeliverySnapshotStatuses,
   AdsHistoricalCampaignRollupStatuses,
@@ -116,6 +117,7 @@ import {
   TrafficEventKinds,
   TrafficSeriesGranularities,
   UNATTRIBUTED_MENTION_REASON,
+  URL_PATH_NORMALIZATION_VERSION,
   USER_PASSWORD_MIN_LENGTH,
   UserRoles,
   adsActivateTreeRequestSchema,
@@ -196,7 +198,7 @@ import {
   visibilityCompareSelectionSchema,
   visibilityReportRequestSchema,
   winnabilityClassSchema
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/cli.ts
 import { pathToFileURL } from "url";
@@ -730,7 +732,7 @@ function parseDemoTrustedProxies(values) {
 async function demoCommand(options) {
   const { host, port } = parseDemoListenOptions(options);
   const trustProxy = parseDemoTrustedProxies(options.trustProxy);
-  const { createDemoServer } = await import("./demo-server-2622GU2H.js");
+  const { createDemoServer } = await import("./demo-server-UGILQVEC.js");
   const app = await createDemoServer({ trustProxy });
   let stopping = false;
   const stop = () => {
@@ -16458,7 +16460,7 @@ function encodeLegacySetupState(state) {
 }
 
 // src/commands/serve.ts
-import { and, asc, eq as eq2, ne } from "drizzle-orm";
+import { and, asc, eq as eq3, ne } from "drizzle-orm";
 
 // src/server-shutdown.ts
 async function closeWithIdleSweep(app, intervalMs = 250) {
@@ -16471,6 +16473,33 @@ async function closeWithIdleSweep(app, intervalMs = 250) {
   }
 }
 
+// src/startup-path-repairs.ts
+import { eq as eq2, lt } from "drizzle-orm";
+var STARTUP_REPAIR_PAGE_SIZE = 4096;
+function repairNormalizedPathsOnStartup(db) {
+  return repairPathsOnce(db, "ga-traffic-paths", backfillNormalizedPaths);
+}
+function repairAiReferralPathsOnStartup(db) {
+  return repairPathsOnce(db, "ga-ai-referral-paths", backfillAiReferralPaths);
+}
+async function repairPathsOnce(db, name, repair) {
+  const completed = db.select({ version: dataRepairCompletions.version }).from(dataRepairCompletions).where(eq2(dataRepairCompletions.name, name)).get();
+  if (completed && completed.version >= URL_PATH_NORMALIZATION_VERSION) return null;
+  let conflicted = false;
+  const result = await repair(db, { pageSize: STARTUP_REPAIR_PAGE_SIZE, onConflict: () => {
+    conflicted = true;
+  } });
+  if (!conflicted) {
+    const completion = { version: URL_PATH_NORMALIZATION_VERSION, completedAt: (/* @__PURE__ */ new Date()).toISOString() };
+    db.insert(dataRepairCompletions).values({ name, ...completion }).onConflictDoUpdate({
+      target: dataRepairCompletions.name,
+      set: completion,
+      setWhere: lt(dataRepairCompletions.version, URL_PATH_NORMALIZATION_VERSION)
+    }).run();
+  }
+  return result;
+}
+
 // src/agent-plugin.ts
 import fs18 from "fs";
 import path12 from "path";
@@ -16683,7 +16712,7 @@ function readServeOpenState(db) {
         projectId: runs.projectId,
         status: runs.status
       }).from(runs).where(and(
-        eq2(runs.kind, RunKinds["site-audit"]),
+        eq3(runs.kind, RunKinds["site-audit"]),
         ne(runs.trigger, RunTriggers.probe)
       )).all();
       scanned = new Set(
@@ -16718,8 +16747,8 @@ async function serveCommand(format = "text") {
   const db = createClient(config.database);
   migrate(db);
   try {
-    const result = backfillNormalizedPaths(db);
-    if (result.updated > 0 && format === "text") {
+    const result = await repairNormalizedPathsOnStartup(db);
+    if (result && result.updated > 0 && format === "text") {
       console.log(
         `Migrated ${result.updated} GA landing-page row${result.updated === 1 ? "" : "s"} to canonical form.`
       );
@@ -16730,8 +16759,8 @@ async function serveCommand(format = "text") {
 `);
   }
   try {
-    const result = backfillAiReferralPaths(db);
-    if (result.updated > 0 && format === "text") {
+    const result = await repairAiReferralPathsOnStartup(db);
+    if (result && result.updated > 0 && format === "text") {
       console.log(
         `Migrated ${result.updated} GA AI referral row${result.updated === 1 ? "" : "s"} to canonical form.`
       );

package/dist/{demo-server-2622GU2H.js → demo-server-UGILQVEC.js}

@@ -5,11 +5,11 @@ import {
   computeFactorAverages,
   layoutSiteCrawlGraphInput,
   persistSiteCrawlGraphLayout
-} from "./chunk-HB6XZOZ6.js";
+} from "./chunk-MSFBNFTI.js";
 import {
   PACKAGE_VERSION,
   canonryMcpTools
-} from "./chunk-6CR7DWLU.js";
+} from "./chunk-5QH72CTN.js";
 import {
   HISTORY_WINDOW_RUNS,
   IntelligenceService,
@@ -85,7 +85,7 @@ import {
   siteCrawlPages,
   siteCrawlSnapshots,
   trafficSources
-} from "./chunk-N7MRAN3G.js";
+} from "./chunk-MTEVU6KG.js";
 import {
   RunKinds,
   SiteCrawlFetchStates,
@@ -112,7 +112,7 @@ import {
   roundPreservingTotal,
   siteAuditPageFactorSchema,
   wilsonInterval
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/demo-server.ts
 import { dirname, join as join2 } from "path";

package/dist/index.js

@@ -3,14 +3,14 @@ import {
   createGoogleMarketingCredentialStore,
   createGoogleMarketingRuntime,
   createServer
-} from "./chunk-5LZLDXEN.js";
+} from "./chunk-4UEX44C3.js";
 import {
   loadConfig
-} from "./chunk-YOMLHHJS.js";
-import "./chunk-HB6XZOZ6.js";
-import "./chunk-6CR7DWLU.js";
-import "./chunk-N7MRAN3G.js";
-import "./chunk-4MWLHTRS.js";
+} from "./chunk-FRBKYQN7.js";
+import "./chunk-MSFBNFTI.js";
+import "./chunk-5QH72CTN.js";
+import "./chunk-MTEVU6KG.js";
+import "./chunk-J7LV4CS2.js";
 export {
   GoogleMarketingRuntimeError,
   createGoogleMarketingCredentialStore,

package/dist/{intelligence-service-QUAOWEIA.js → intelligence-service-JYGP6LME.js}

@@ -1,8 +1,8 @@
 import {
   HISTORY_WINDOW_RUNS,
   IntelligenceService
-} from "./chunk-N7MRAN3G.js";
-import "./chunk-4MWLHTRS.js";
+} from "./chunk-MTEVU6KG.js";
+import "./chunk-J7LV4CS2.js";
 export {
   HISTORY_WINDOW_RUNS,
   IntelligenceService

package/dist/mcp.js

@@ -1,15 +1,15 @@
 import {
   autoSyncSkills
-} from "./chunk-TBAVVIVI.js";
+} from "./chunk-PAYC6TQ2.js";
 import {
   createApiClient,
   createCanonryMcpServer
-} from "./chunk-YOMLHHJS.js";
-import "./chunk-6CR7DWLU.js";
+} from "./chunk-FRBKYQN7.js";
+import "./chunk-5QH72CTN.js";
 import {
   isReadOnlyKey,
   updateCheckEnvOptOut
-} from "./chunk-4MWLHTRS.js";
+} from "./chunk-J7LV4CS2.js";
 
 // src/mcp/cli.ts
 import { randomUUID } from "crypto";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@canonry/canonry",
-  "version": "6.1.4",
+  "version": "6.1.5",
   "type": "module",
   "description": "Self-hosted AI visibility (AEO) platform: track how ChatGPT, Claude, Gemini, and Perplexity cite your domain, join it with Search Console, GA4, server-side traffic, and paid media, and fix what you find through agent tools (CLI, REST, MCP). Local SQLite.",
   "keywords": [
@@ -89,27 +89,27 @@
     "@ainyc/canonry-api-client": "0.0.0",
     "@ainyc/canonry-api-routes": "0.0.0",
     "@ainyc/canonry-config": "0.0.0",
-    "@ainyc/canonry-contracts": "0.0.0",
     "@ainyc/canonry-db": "0.0.0",
+    "@ainyc/canonry-contracts": "0.0.0",
     "@ainyc/canonry-integration-bing": "0.0.0",
     "@ainyc/canonry-integration-cloud-run": "0.0.0",
     "@ainyc/canonry-integration-cloudflare-queue": "0.0.0",
-    "@ainyc/canonry-integration-cloudflare-worker": "0.0.0",
     "@ainyc/canonry-integration-commoncrawl": "0.0.0",
+    "@ainyc/canonry-integration-cloudflare-worker": "0.0.0",
     "@ainyc/canonry-integration-google": "0.0.0",
-    "@ainyc/canonry-integration-google-ads": "0.0.0",
-    "@ainyc/canonry-integration-google-places": "0.0.0",
     "@ainyc/canonry-integration-google-business-profile": "0.0.0",
-    "@ainyc/canonry-integration-openai-ads": "0.0.0",
+    "@ainyc/canonry-integration-google-ads": "0.0.0",
     "@ainyc/canonry-integration-google-tag-manager": "0.0.0",
+    "@ainyc/canonry-integration-openai-ads": "0.0.0",
     "@ainyc/canonry-integration-traffic": "0.0.0",
+    "@ainyc/canonry-intelligence": "0.0.0",
     "@ainyc/canonry-integration-wordpress": "0.0.0",
-    "@ainyc/canonry-provider-cdp": "0.0.0",
+    "@ainyc/canonry-integration-google-places": "0.0.0",
     "@ainyc/canonry-provider-claude": "0.0.0",
-    "@ainyc/canonry-intelligence": "0.0.0",
+    "@ainyc/canonry-provider-cdp": "0.0.0",
     "@ainyc/canonry-provider-gemini": "0.0.0",
-    "@ainyc/canonry-provider-local": "0.0.0",
     "@ainyc/canonry-provider-muse": "0.0.0",
+    "@ainyc/canonry-provider-local": "0.0.0",
     "@ainyc/canonry-provider-perplexity": "0.0.0",
     "@ainyc/canonry-integration-typesafe": "0.0.0",
     "@ainyc/canonry-provider-openai": "0.0.0"