@ainyc/canonry 4.15.0 → 4.15.2

package/dist/{chunk-DLSQXNUN.js → chunk-IVNWS2YU.js}

@@ -5,7 +5,7 @@ import {
   loadConfig,
   loadConfigRaw,
   saveConfigPatch
-} from "./chunk-C32VL5BB.js";
+} from "./chunk-7SRKUAZO.js";
 import {
   DEFAULT_RUN_HISTORY_LIMIT,
   IntelligenceService,
@@ -66,7 +66,7 @@ import {
   schedules,
   trafficSources,
   usageCounters
-} from "./chunk-7HBZCGRL.js";
+} from "./chunk-MI33SQL6.js";
 import {
   AGENT_MEMORY_VALUE_MAX_BYTES,
   AGENT_PROVIDER_IDS,
@@ -82,6 +82,7 @@ import {
   RunStatuses,
   RunTriggers,
   TrafficEventConfidences,
+  TrafficEventKinds,
   TrafficEvidenceKinds,
   TrafficSourceAuthModes,
   TrafficSourceStatuses,
@@ -152,7 +153,7 @@ import {
   visibilityStateFromAnswerMentioned,
   windowCutoff,
   wordpressEnvSchema
-} from "./chunk-6QTH5NS5.js";
+} from "./chunk-ONI3TX2A.js";
 
 // src/telemetry.ts
 import crypto from "crypto";
@@ -9661,8 +9662,66 @@ var routeCatalog = [
     },
     responses: {
       200: { description: "Sync summary returned." },
-      400: { description: "Invalid sync request, missing credentials, or upstream pull error." },
-      404: { description: "Project or traffic source not found." }
+      400: { description: "Invalid sync request or missing credentials." },
+      404: { description: "Project or traffic source not found." },
+      502: { description: "Upstream Cloud Run pull or auth-token resolution failed." }
+    }
+  },
+  {
+    method: "get",
+    path: "/api/v1/projects/{name}/traffic/sources",
+    summary: "List non-archived traffic sources for a project",
+    tags: ["traffic"],
+    parameters: [nameParameter],
+    responses: {
+      200: { description: "Source list returned." },
+      404: { description: "Project not found." }
+    }
+  },
+  {
+    method: "get",
+    path: "/api/v1/projects/{name}/traffic/status",
+    summary: "List non-archived traffic sources with last-24h totals and the latest sync run for each",
+    description: "Single-call composite for the `canonry traffic status` view: same shape as `GET /traffic/sources/{id}` but returned as `{ sources: TrafficSourceDetailDto[] }` for every non-archived source. Lets agents and the dashboard avoid an N+1 fan-out.",
+    tags: ["traffic"],
+    parameters: [nameParameter],
+    responses: {
+      200: { description: "Status returned." },
+      404: { description: "Project not found." }
+    }
+  },
+  {
+    method: "get",
+    path: "/api/v1/projects/{name}/traffic/sources/{id}",
+    summary: "Get a single traffic source with last-24h totals and the latest sync run",
+    tags: ["traffic"],
+    parameters: [
+      nameParameter,
+      { name: "id", in: "path", required: true, description: "Traffic source ID.", schema: stringSchema }
+    ],
+    responses: {
+      200: { description: "Source detail returned." },
+      404: { description: "Project or source not found." }
+    }
+  },
+  {
+    method: "get",
+    path: "/api/v1/projects/{name}/traffic/events",
+    summary: "List rolled-up crawler and AI-referral hits within a window",
+    description: "Returns hourly rollup rows from `crawler_events_hourly` and `ai_referral_events_hourly`. Defaults to the last 24h. Totals reflect the full window; the `events` array is capped by `limit` (default 500, max 5000).",
+    tags: ["traffic"],
+    parameters: [
+      nameParameter,
+      { name: "since", in: "query", description: "ISO-8601 window start (defaults to 24h ago).", schema: stringSchema },
+      { name: "until", in: "query", description: "ISO-8601 window end (defaults to now).", schema: stringSchema },
+      { name: "kind", in: "query", description: 'Filter to "crawler", "ai-referral", or "all" (default).', schema: stringSchema },
+      { name: "limit", in: "query", description: "Max rows per kind in the events array (default 500, max 5000).", schema: stringSchema },
+      { name: "sourceId", in: "query", description: "Restrict to a single traffic source.", schema: stringSchema }
+    ],
+    responses: {
+      200: { description: "Events returned with windowed totals." },
+      400: { description: "Invalid query parameters." },
+      404: { description: "Project not found." }
     }
   }
 ];
@@ -15893,7 +15952,7 @@ async function backlinksRoutes(app, opts) {
 
 // ../api-routes/src/traffic.ts
 import crypto20 from "crypto";
-import { eq as eq23, sql as sql7 } from "drizzle-orm";
+import { and as and12, desc as desc12, eq as eq23, gte, lte, sql as sql7 } from "drizzle-orm";
 
 // ../integration-cloud-run/src/auth.ts
 import crypto19 from "crypto";
@@ -16500,10 +16559,11 @@ function incrementBucket(map, key, fields) {
 }
 
 // ../api-routes/src/traffic.ts
-var DEFAULT_SYNC_WINDOW_MINUTES = 60;
+var DEFAULT_SYNC_WINDOW_MINUTES = 43200;
 var DEFAULT_PAGE_SIZE2 = 1e3;
 var DEFAULT_MAX_PAGES2 = 5;
 var DEFAULT_SAMPLE_LIMIT2 = 100;
+var MAX_TRACKED_EVENT_IDS = 1e3;
 function parseSourceConfig(row) {
   return parseJsonColumn(row.configJson, {});
 }
@@ -16664,17 +16724,24 @@ async function trafficRoutes(app, opts) {
       kind: RunKinds["traffic-sync"],
       status: RunStatuses.running,
       trigger: RunTriggers.manual,
+      sourceId: sourceRow.id,
       startedAt,
       createdAt: startedAt
     }).run();
+    const markFailed = (msg) => {
+      const failedAt = (/* @__PURE__ */ new Date()).toISOString();
+      app.db.transaction((tx) => {
+        tx.update(runs).set({ status: RunStatuses.failed, error: msg, finishedAt: failedAt }).where(eq23(runs.id, runId)).run();
+        tx.update(trafficSources).set({ status: TrafficSourceStatuses.error, lastError: msg, updatedAt: failedAt }).where(eq23(trafficSources.id, sourceRow.id)).run();
+      });
+    };
     let accessToken;
     try {
       accessToken = await resolveAccessToken2(credential);
     } catch (e) {
       const msg = e instanceof Error ? e.message : String(e);
-      app.db.update(runs).set({ status: RunStatuses.failed, error: msg, finishedAt: (/* @__PURE__ */ new Date()).toISOString() }).where(eq23(runs.id, runId)).run();
-      app.db.update(trafficSources).set({ status: TrafficSourceStatuses.error, lastError: msg, updatedAt: (/* @__PURE__ */ new Date()).toISOString() }).where(eq23(trafficSources.id, sourceRow.id)).run();
-      throw validationError(`Failed to resolve Cloud Run access token: ${msg}`);
+      markFailed(msg);
+      throw providerError(`Failed to resolve Cloud Run access token: ${msg}`);
     }
     let allEvents = [];
     try {
@@ -16690,11 +16757,23 @@ async function trafficRoutes(app, opts) {
       allEvents = page.events;
     } catch (e) {
       const msg = e instanceof Error ? e.message : String(e);
-      app.db.update(runs).set({ status: RunStatuses.failed, error: msg, finishedAt: (/* @__PURE__ */ new Date()).toISOString() }).where(eq23(runs.id, runId)).run();
-      app.db.update(trafficSources).set({ status: TrafficSourceStatuses.error, lastError: msg, updatedAt: (/* @__PURE__ */ new Date()).toISOString() }).where(eq23(trafficSources.id, sourceRow.id)).run();
-      throw validationError(`Cloud Run pull failed: ${msg}`);
-    }
-    const report = buildTrafficProbeReport(allEvents, { sampleLimit });
+      markFailed(msg);
+      throw providerError(`Cloud Run pull failed: ${msg}`);
+    }
+    const seenEventIds = new Set(parseJsonColumn(sourceRow.lastEventIds, []));
+    const dedupedEvents = seenEventIds.size === 0 ? allEvents : allEvents.filter((e) => !seenEventIds.has(e.eventId));
+    const newSorted = dedupedEvents.slice().sort((a, b) => a.observedAt < b.observedAt ? 1 : a.observedAt > b.observedAt ? -1 : 0).map((e) => e.eventId);
+    const previousIds = parseJsonColumn(sourceRow.lastEventIds, []);
+    const merged = [];
+    const mergedSet = /* @__PURE__ */ new Set();
+    for (const id of [...newSorted, ...previousIds]) {
+      if (mergedSet.has(id)) continue;
+      mergedSet.add(id);
+      merged.push(id);
+      if (merged.length >= MAX_TRACKED_EVENT_IDS) break;
+    }
+    const nextEventIds = merged;
+    const report = buildTrafficProbeReport(dedupedEvents, { sampleLimit });
     const finishedAt = (/* @__PURE__ */ new Date()).toISOString();
     let crawlerBucketRows = 0;
     let aiReferralBucketRows = 0;
@@ -16800,6 +16879,7 @@ async function trafficRoutes(app, opts) {
         status: TrafficSourceStatuses.connected,
         lastSyncedAt: finishedAt,
         lastError: null,
+        lastEventIds: JSON.stringify(nextEventIds),
         updatedAt: finishedAt
       }).where(eq23(trafficSources.id, sourceRow.id)).run();
       tx.update(runs).set({ status: RunStatuses.completed, finishedAt }).where(eq23(runs.id, runId)).run();
@@ -16827,6 +16907,177 @@ async function trafficRoutes(app, opts) {
     };
     return response;
   });
+  function buildSourceDetail(projectId, row, since) {
+    const crawlerTotals = app.db.select({ total: sql7`COALESCE(SUM(${crawlerEventsHourly.hits}), 0)` }).from(crawlerEventsHourly).where(
+      and12(
+        eq23(crawlerEventsHourly.sourceId, row.id),
+        gte(crawlerEventsHourly.tsHour, since)
+      )
+    ).get();
+    const aiTotals = app.db.select({ total: sql7`COALESCE(SUM(${aiReferralEventsHourly.sessionsOrHits}), 0)` }).from(aiReferralEventsHourly).where(
+      and12(
+        eq23(aiReferralEventsHourly.sourceId, row.id),
+        gte(aiReferralEventsHourly.tsHour, since)
+      )
+    ).get();
+    const sampleTotals = app.db.select({ total: sql7`COUNT(*)` }).from(rawEventSamples).where(
+      and12(
+        eq23(rawEventSamples.sourceId, row.id),
+        gte(rawEventSamples.ts, since)
+      )
+    ).get();
+    const latestRun = app.db.select().from(runs).where(
+      and12(
+        eq23(runs.projectId, projectId),
+        eq23(runs.kind, RunKinds["traffic-sync"]),
+        eq23(runs.sourceId, row.id)
+      )
+    ).orderBy(desc12(runs.startedAt)).limit(1).get();
+    return {
+      ...rowToDto(row),
+      totals24h: {
+        crawlerHits: Number(crawlerTotals?.total ?? 0),
+        aiReferralHits: Number(aiTotals?.total ?? 0),
+        sampleCount: Number(sampleTotals?.total ?? 0)
+      },
+      latestRun: latestRun ? {
+        runId: latestRun.id,
+        status: latestRun.status,
+        startedAt: latestRun.startedAt,
+        finishedAt: latestRun.finishedAt ?? null,
+        error: latestRun.error ?? null
+      } : null
+    };
+  }
+  app.get("/projects/:name/traffic/sources", async (request) => {
+    const project = resolveProject(app.db, request.params.name);
+    const rows = app.db.select().from(trafficSources).where(eq23(trafficSources.projectId, project.id)).orderBy(desc12(trafficSources.createdAt)).all();
+    const sources = rows.filter((row) => row.status !== TrafficSourceStatuses.archived).map(rowToDto);
+    const response = { sources };
+    return response;
+  });
+  app.get("/projects/:name/traffic/status", async (request) => {
+    const project = resolveProject(app.db, request.params.name);
+    const rows = app.db.select().from(trafficSources).where(eq23(trafficSources.projectId, project.id)).orderBy(desc12(trafficSources.createdAt)).all();
+    const since = new Date(Date.now() - 24 * 60 * 6e4).toISOString();
+    const sources = rows.filter((row) => row.status !== TrafficSourceStatuses.archived).map((row) => buildSourceDetail(project.id, row, since));
+    const response = { sources };
+    return response;
+  });
+  app.get(
+    "/projects/:name/traffic/sources/:id",
+    async (request) => {
+      const project = resolveProject(app.db, request.params.name);
+      const row = app.db.select().from(trafficSources).where(eq23(trafficSources.id, request.params.id)).get();
+      if (!row || row.projectId !== project.id) {
+        throw notFound("Traffic source", request.params.id);
+      }
+      const since = new Date(Date.now() - 24 * 60 * 6e4).toISOString();
+      return buildSourceDetail(project.id, row, since);
+    }
+  );
+  app.get("/projects/:name/traffic/events", async (request) => {
+    const project = resolveProject(app.db, request.params.name);
+    const now = /* @__PURE__ */ new Date();
+    const defaultSince = new Date(now.getTime() - 24 * 60 * 6e4);
+    const sinceParam = request.query?.since;
+    const untilParam = request.query?.until;
+    const since = sinceParam ? new Date(sinceParam) : defaultSince;
+    const until = untilParam ? new Date(untilParam) : now;
+    if (Number.isNaN(since.getTime())) {
+      throw validationError('"since" must be an ISO-8601 timestamp');
+    }
+    if (Number.isNaN(until.getTime())) {
+      throw validationError('"until" must be an ISO-8601 timestamp');
+    }
+    if (since.getTime() > until.getTime()) {
+      throw validationError('"since" must be \u2264 "until"');
+    }
+    const kindParam = request.query?.kind;
+    let kind = "all";
+    if (kindParam !== void 0) {
+      if (kindParam === "all" || kindParam === TrafficEventKinds.crawler || kindParam === TrafficEventKinds["ai-referral"]) {
+        kind = kindParam;
+      } else {
+        throw validationError(`"kind" must be one of: all, ${TrafficEventKinds.crawler}, ${TrafficEventKinds["ai-referral"]}`);
+      }
+    }
+    const limitParam = request.query?.limit;
+    const requestedLimit = limitParam ? parseInt(limitParam, 10) : 500;
+    if (!Number.isFinite(requestedLimit) || requestedLimit <= 0) {
+      throw validationError('"limit" must be a positive integer');
+    }
+    const limit = Math.min(requestedLimit, 5e3);
+    const sourceIdParam = request.query?.sourceId;
+    const sinceIso = since.toISOString();
+    const untilIso = until.toISOString();
+    const events = [];
+    let crawlerTotal = 0;
+    let aiReferralTotal = 0;
+    if (kind === "all" || kind === TrafficEventKinds.crawler) {
+      const crawlerFilters = [
+        eq23(crawlerEventsHourly.projectId, project.id),
+        gte(crawlerEventsHourly.tsHour, sinceIso),
+        lte(crawlerEventsHourly.tsHour, untilIso)
+      ];
+      if (sourceIdParam) crawlerFilters.push(eq23(crawlerEventsHourly.sourceId, sourceIdParam));
+      const crawlerWhere = and12(...crawlerFilters);
+      const total = app.db.select({ total: sql7`COALESCE(SUM(${crawlerEventsHourly.hits}), 0)` }).from(crawlerEventsHourly).where(crawlerWhere).get();
+      crawlerTotal = Number(total?.total ?? 0);
+      const rows = app.db.select().from(crawlerEventsHourly).where(crawlerWhere).orderBy(desc12(crawlerEventsHourly.tsHour)).limit(limit).all();
+      for (const r of rows) {
+        events.push({
+          kind: TrafficEventKinds.crawler,
+          sourceId: r.sourceId,
+          tsHour: r.tsHour,
+          botId: r.botId,
+          operator: r.operator,
+          verificationStatus: r.verificationStatus,
+          pathNormalized: r.pathNormalized,
+          status: r.status,
+          hits: r.hits
+        });
+      }
+    }
+    if (kind === "all" || kind === TrafficEventKinds["ai-referral"]) {
+      const aiFilters = [
+        eq23(aiReferralEventsHourly.projectId, project.id),
+        gte(aiReferralEventsHourly.tsHour, sinceIso),
+        lte(aiReferralEventsHourly.tsHour, untilIso)
+      ];
+      if (sourceIdParam) aiFilters.push(eq23(aiReferralEventsHourly.sourceId, sourceIdParam));
+      const aiWhere = and12(...aiFilters);
+      const total = app.db.select({ total: sql7`COALESCE(SUM(${aiReferralEventsHourly.sessionsOrHits}), 0)` }).from(aiReferralEventsHourly).where(aiWhere).get();
+      aiReferralTotal = Number(total?.total ?? 0);
+      const rows = app.db.select().from(aiReferralEventsHourly).where(aiWhere).orderBy(desc12(aiReferralEventsHourly.tsHour)).limit(limit).all();
+      for (const r of rows) {
+        events.push({
+          kind: TrafficEventKinds["ai-referral"],
+          sourceId: r.sourceId,
+          tsHour: r.tsHour,
+          product: r.product,
+          operator: r.operator,
+          sourceDomain: r.sourceDomain,
+          evidenceType: r.evidenceType,
+          landingPathNormalized: r.landingPathNormalized,
+          status: r.status,
+          hits: r.sessionsOrHits
+        });
+      }
+    }
+    events.sort((a, b) => a.tsHour < b.tsHour ? 1 : a.tsHour > b.tsHour ? -1 : 0);
+    const trimmed = events.slice(0, limit);
+    const response = {
+      windowStart: sinceIso,
+      windowEnd: untilIso,
+      totals: {
+        crawlerHits: crawlerTotal,
+        aiReferralHits: aiReferralTotal
+      },
+      events: trimmed
+    };
+    return response;
+  });
 }
 
 // ../api-routes/src/doctor/checks/bing-auth.ts
@@ -20198,7 +20449,7 @@ import crypto22 from "crypto";
 import fs7 from "fs";
 import path9 from "path";
 import os5 from "os";
-import { and as and12, eq as eq24, inArray as inArray7, sql as sql8 } from "drizzle-orm";
+import { and as and13, eq as eq24, inArray as inArray7, sql as sql8 } from "drizzle-orm";
 
 // src/run-telemetry.ts
 import crypto21 from "crypto";
@@ -20577,7 +20828,7 @@ var JobRunner = class {
         throw new Error(`Run ${runId} is not executable from status '${existingRun.status}'`);
       }
       if (existingRun.status === "queued") {
-        this.db.update(runs).set({ status: "running", startedAt: now }).where(and12(eq24(runs.id, runId), eq24(runs.status, "queued"))).run();
+        this.db.update(runs).set({ status: "running", startedAt: now }).where(and13(eq24(runs.id, runId), eq24(runs.status, "queued"))).run();
       }
       this.throwIfRunCancelled(runId);
       const project = this.db.select().from(projects).where(eq24(projects.id, projectId)).get();
@@ -20936,7 +21187,7 @@ function buildPhases(input) {
 
 // src/gsc-sync.ts
 import crypto23 from "crypto";
-import { eq as eq25, and as and13, sql as sql9 } from "drizzle-orm";
+import { eq as eq25, and as and14, sql as sql9 } from "drizzle-orm";
 var log2 = createLogger("GscSync");
 function formatDate3(d) {
   return d.toISOString().split("T")[0];
@@ -20988,7 +21239,7 @@ async function executeGscSync(db, runId, projectId, opts) {
     });
     log2.info("fetch.complete", { runId, projectId, rowCount: rows.length });
     db.delete(gscSearchData).where(
-      and13(
+      and14(
         eq25(gscSearchData.projectId, projectId),
         sql9`${gscSearchData.date} >= ${startDate}`,
         sql9`${gscSearchData.date} <= ${endDate}`
@@ -21077,7 +21328,7 @@ async function executeGscSync(db, runId, projectId, opts) {
       }
     }
     const snapshotDate = formatDate3(/* @__PURE__ */ new Date());
-    db.delete(gscCoverageSnapshots).where(and13(eq25(gscCoverageSnapshots.projectId, projectId), eq25(gscCoverageSnapshots.date, snapshotDate))).run();
+    db.delete(gscCoverageSnapshots).where(and14(eq25(gscCoverageSnapshots.projectId, projectId), eq25(gscCoverageSnapshots.date, snapshotDate))).run();
     db.insert(gscCoverageSnapshots).values({
       id: crypto23.randomUUID(),
       projectId,
@@ -21100,7 +21351,7 @@ async function executeGscSync(db, runId, projectId, opts) {
 
 // src/gsc-inspect-sitemap.ts
 import crypto24 from "crypto";
-import { eq as eq26, and as and14 } from "drizzle-orm";
+import { eq as eq26, and as and15 } from "drizzle-orm";
 
 // src/sitemap-parser.ts
 var log3 = createLogger("SitemapParser");
@@ -21316,7 +21567,7 @@ async function executeInspectSitemap(db, runId, projectId, opts) {
       }
     }
     const snapshotDate = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
-    db.delete(gscCoverageSnapshots).where(and14(eq26(gscCoverageSnapshots.projectId, projectId), eq26(gscCoverageSnapshots.date, snapshotDate))).run();
+    db.delete(gscCoverageSnapshots).where(and15(eq26(gscCoverageSnapshots.projectId, projectId), eq26(gscCoverageSnapshots.date, snapshotDate))).run();
     db.insert(gscCoverageSnapshots).values({
       id: crypto24.randomUUID(),
       projectId,
@@ -21340,7 +21591,7 @@ async function executeInspectSitemap(db, runId, projectId, opts) {
 
 // src/bing-inspect-sitemap.ts
 import crypto25 from "crypto";
-import { eq as eq27, desc as desc12 } from "drizzle-orm";
+import { eq as eq27, desc as desc13 } from "drizzle-orm";
 var log5 = createLogger("BingInspectSitemap");
 function parseBingDate2(value) {
   if (!value) return null;
@@ -21461,7 +21712,7 @@ async function executeBingInspectSitemap(db, runId, projectId, opts) {
         await new Promise((r) => setTimeout(r, 1e3));
       }
     }
-    const allInspections = db.select().from(bingUrlInspections).where(eq27(bingUrlInspections.projectId, projectId)).orderBy(desc12(bingUrlInspections.inspectedAt)).all();
+    const allInspections = db.select().from(bingUrlInspections).where(eq27(bingUrlInspections.projectId, projectId)).orderBy(desc13(bingUrlInspections.inspectedAt)).all();
     const latestByUrl = /* @__PURE__ */ new Map();
     const definitiveByUrl = /* @__PURE__ */ new Map();
     for (const row of allInspections) {
@@ -21527,7 +21778,7 @@ async function executeBingInspectSitemap(db, runId, projectId, opts) {
 // src/commoncrawl-sync.ts
 import crypto26 from "crypto";
 import path10 from "path";
-import { and as and15, eq as eq28, sql as sql10 } from "drizzle-orm";
+import { and as and16, eq as eq28, sql as sql10 } from "drizzle-orm";
 var log6 = createLogger("CommonCrawlSync");
 var INSERT_CHUNK_SIZE = 1e4;
 function defaultDeps() {
@@ -21718,7 +21969,7 @@ function computeSummary(rows) {
 // src/backlink-extract.ts
 import crypto27 from "crypto";
 import fs8 from "fs";
-import { and as and16, desc as desc13, eq as eq29 } from "drizzle-orm";
+import { and as and17, desc as desc14, eq as eq29 } from "drizzle-orm";
 var log7 = createLogger("BacklinkExtract");
 function defaultDeps2() {
   return {
@@ -21736,7 +21987,7 @@ async function executeBacklinkExtract(db, runId, projectId, opts = {}) {
     if (!project) {
       throw new Error(`Project not found: ${projectId}`);
     }
-    const sync = opts.release ? db.select().from(ccReleaseSyncs).where(eq29(ccReleaseSyncs.release, opts.release)).get() : db.select().from(ccReleaseSyncs).where(eq29(ccReleaseSyncs.status, CcReleaseSyncStatuses.ready)).orderBy(desc13(ccReleaseSyncs.createdAt)).limit(1).get();
+    const sync = opts.release ? db.select().from(ccReleaseSyncs).where(eq29(ccReleaseSyncs.release, opts.release)).get() : db.select().from(ccReleaseSyncs).where(eq29(ccReleaseSyncs.status, CcReleaseSyncStatuses.ready)).orderBy(desc14(ccReleaseSyncs.createdAt)).limit(1).get();
     if (!sync) {
       throw new Error("No ready release sync available \u2014 run `canonry backlinks sync` first");
     }
@@ -21764,7 +22015,7 @@ async function executeBacklinkExtract(db, runId, projectId, opts = {}) {
     const targetDomain = project.canonicalDomain;
     db.transaction((tx) => {
       tx.delete(backlinkDomains).where(
-        and16(eq29(backlinkDomains.projectId, projectId), eq29(backlinkDomains.release, release))
+        and17(eq29(backlinkDomains.projectId, projectId), eq29(backlinkDomains.release, release))
       ).run();
       if (rows.length > 0) {
         const values = rows.map((r) => ({
@@ -22020,7 +22271,7 @@ var Scheduler = class {
 };
 
 // src/notifier.ts
-import { eq as eq31, desc as desc14, and as and17, or as or4 } from "drizzle-orm";
+import { eq as eq31, desc as desc15, and as and18, or as or4 } from "drizzle-orm";
 import crypto28 from "crypto";
 var log9 = createLogger("Notifier");
 var Notifier = class {
@@ -22126,11 +22377,11 @@ var Notifier = class {
   }
   computeTransitions(runId, projectId) {
     const recentRuns = this.db.select().from(runs).where(
-      and17(
+      and18(
         eq31(runs.projectId, projectId),
         or4(eq31(runs.status, "completed"), eq31(runs.status, "partial"))
       )
-    ).orderBy(desc14(runs.createdAt)).limit(2).all();
+    ).orderBy(desc15(runs.createdAt)).limit(2).all();
     if (recentRuns.length < 2) return [];
     const currentRunId = recentRuns[0].id;
     const previousRunId = recentRuns[1].id;
@@ -22612,7 +22863,7 @@ function resolveSessionProviderAndModel(config, opts) {
 
 // src/agent/memory-store.ts
 import crypto29 from "crypto";
-import { and as and18, desc as desc15, eq as eq32, like as like2, sql as sql11 } from "drizzle-orm";
+import { and as and19, desc as desc16, eq as eq32, like as like2, sql as sql11 } from "drizzle-orm";
 var COMPACTION_KEY_PREFIX = "compaction:";
 var COMPACTION_NOTES_PER_SESSION = 3;
 function rowToDto2(row) {
@@ -22626,7 +22877,7 @@ function rowToDto2(row) {
   };
 }
 function listMemoryEntries(db, projectId, opts = {}) {
-  const query = db.select().from(agentMemory).where(eq32(agentMemory.projectId, projectId)).orderBy(desc15(agentMemory.updatedAt));
+  const query = db.select().from(agentMemory).where(eq32(agentMemory.projectId, projectId)).orderBy(desc16(agentMemory.updatedAt));
   const rows = opts.limit === void 0 ? query.all() : query.limit(opts.limit).all();
   return rows.map(rowToDto2);
 }
@@ -22657,12 +22908,12 @@ function upsertMemoryEntry(db, args) {
       updatedAt: now
     }
   }).run();
-  const row = db.select().from(agentMemory).where(and18(eq32(agentMemory.projectId, args.projectId), eq32(agentMemory.key, args.key))).get();
+  const row = db.select().from(agentMemory).where(and19(eq32(agentMemory.projectId, args.projectId), eq32(agentMemory.key, args.key))).get();
   if (!row) throw new Error("memory upsert produced no row");
   return rowToDto2(row);
 }
 function deleteMemoryEntry(db, projectId, key) {
-  const result = db.delete(agentMemory).where(and18(eq32(agentMemory.projectId, projectId), eq32(agentMemory.key, key))).run();
+  const result = db.delete(agentMemory).where(and19(eq32(agentMemory.projectId, projectId), eq32(agentMemory.key, key))).run();
   const changes = result.changes ?? 0;
   return changes > 0;
 }
@@ -22691,16 +22942,16 @@ function writeCompactionNote(db, args) {
     }).run();
     const sessionPrefix = `${COMPACTION_KEY_PREFIX}${args.sessionId}:`;
     const existing = tx.select({ id: agentMemory.id, updatedAt: agentMemory.updatedAt }).from(agentMemory).where(
-      and18(
+      and19(
         eq32(agentMemory.projectId, args.projectId),
         like2(agentMemory.key, `${sessionPrefix}%`)
       )
-    ).orderBy(desc15(agentMemory.updatedAt)).all();
+    ).orderBy(desc16(agentMemory.updatedAt)).all();
     const stale = existing.slice(COMPACTION_NOTES_PER_SESSION).map((r) => r.id);
     if (stale.length > 0) {
       tx.delete(agentMemory).where(sql11`${agentMemory.id} IN (${sql11.join(stale.map((s) => sql11`${s}`), sql11`, `)})`).run();
     }
-    const row = tx.select().from(agentMemory).where(and18(eq32(agentMemory.projectId, args.projectId), eq32(agentMemory.key, key))).get();
+    const row = tx.select().from(agentMemory).where(and19(eq32(agentMemory.projectId, args.projectId), eq32(agentMemory.key, key))).get();
     if (row) inserted = rowToDto2(row);
   });
   if (!inserted) throw new Error("compaction note write produced no row");

package/dist/{chunk-7HBZCGRL.js → chunk-MI33SQL6.js}

@@ -8,7 +8,7 @@ import {
   categoryLabel,
   determineAnswerMentioned,
   normalizeProjectDomain
-} from "./chunk-6QTH5NS5.js";
+} from "./chunk-ONI3TX2A.js";
 
 // src/intelligence-service.ts
 import { eq, desc, asc, and, or, inArray } from "drizzle-orm";
@@ -104,13 +104,15 @@ var runs = sqliteTable("runs", {
   status: text("status").notNull().default("queued"),
   trigger: text("trigger").notNull().default("manual"),
   location: text("location"),
+  sourceId: text("source_id"),
   startedAt: text("started_at"),
   finishedAt: text("finished_at"),
   error: text("error"),
   createdAt: text("created_at").notNull()
 }, (table) => [
   index("idx_runs_project").on(table.projectId),
-  index("idx_runs_status").on(table.status)
+  index("idx_runs_status").on(table.status),
+  index("idx_runs_source").on(table.sourceId)
 ]);
 var querySnapshots = sqliteTable("query_snapshots", {
   id: text("id").primaryKey(),
@@ -555,6 +557,11 @@ var trafficSources = sqliteTable("traffic_sources", {
   lastSyncedAt: text("last_synced_at"),
   lastCursor: text("last_cursor"),
   lastError: text("last_error"),
+  // JSON-encoded array of normalized event IDs (e.g. `cloud-run:<ts>:<insertId>`)
+  // observed in the most recent successful sync. Bounded ring buffer used to
+  // dedupe across sync runs at the boundary timestamp where lastSyncedAt
+  // clamping alone leaves a small overlap window.
+  lastEventIds: text("last_event_ids"),
   archivedAt: text("archived_at"),
   configJson: text("config_json").notNull().default("{}"),
   createdAt: text("created_at").notNull(),
@@ -1586,6 +1593,24 @@ var MIGRATION_VERSIONS = [
       tx.run(sql.raw(`CREATE UNIQUE INDEX IF NOT EXISTS idx_ga_ai_ref_unique_v4
          ON ga_ai_referrals(project_id, date, source, medium, source_dimension, channel_group, landing_page)`));
     }
+  },
+  {
+    version: 51,
+    name: "runs-source-id",
+    statements: [
+      `ALTER TABLE runs ADD COLUMN source_id TEXT`,
+      `CREATE INDEX IF NOT EXISTS idx_runs_source ON runs(source_id)`
+    ]
+  },
+  {
+    version: 52,
+    name: "traffic-sources-last-event-ids",
+    statements: [
+      // JSON-encoded array of normalized event IDs from the previous sync,
+      // used for cross-sync boundary-window dedupe so a longer default
+      // sync window (or any overlapping re-sync) cannot double-count.
+      `ALTER TABLE traffic_sources ADD COLUMN last_event_ids TEXT`
+    ]
   }
 ];
 function isDuplicateColumnError(err) {