@canonry/canonry 7.3.0 → 7.5.0

package/dist/{chunk-RVELNJAW.js → chunk-6Z5MRX24.js}

@@ -5,7 +5,7 @@ import {
   PACKAGE_VERSION,
   canonryMcpTools,
   isMachineFormat
-} from "./chunk-FWV6IIYY.js";
+} from "./chunk-JURAGZOO.js";
 import {
   AGENT_PROVIDER_IDS,
   CANONRY_NPM_PACKAGE_URL,
@@ -50,7 +50,7 @@ import {
   skillsClientSchema,
   upgradeCaveatFor,
   upgradeCommandFor
-} from "./chunk-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/config.ts
 import fs from "fs";

package/dist/{chunk-3JEUZ2K5.js → chunk-E7EWAAH3.js}

@@ -4,14 +4,14 @@ import {
   installSkills,
   loadConfigRaw,
   saveConfigPatch
-} from "./chunk-RVELNJAW.js";
+} from "./chunk-6Z5MRX24.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-FWV6IIYY.js";
+} from "./chunk-JURAGZOO.js";
 import {
   SKILL_MANIFEST_FILENAME,
   SkillsClients
-} from "./chunk-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/skills-autosync.ts
 import fs from "fs";

package/dist/{chunk-FWV6IIYY.js → chunk-JURAGZOO.js}

@@ -124,6 +124,7 @@ import {
   sentimentSummaryIncludeSchema,
   sentimentSummaryReadSchema,
   settingsDtoSchema,
+  siteAuditPageBudgetSchema,
   snapshotRequestInputSchema,
   telemetryStatusDtoSchema,
   trafficAnalyticsPeriodSchema,
@@ -134,7 +135,7 @@ import {
   trafficSeriesGranularitySchema,
   visibilityCompareSelectionSchema,
   visibilityReportRequestSchema
-} from "./chunk-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/cli-error.ts
 function isMachineFormat(format) {
@@ -1209,7 +1210,7 @@ var technicalAeoRunInputSchema = z3.object({
   project: projectNameSchema,
   sitemapUrl: z3.string().url().optional().describe("Override the sitemap URL. Defaults to https://<canonicalDomain>/sitemap.xml."),
   limit: z3.number().int().positive().max(2e3).optional().describe("Deprecated compatibility alias for maxPages."),
-  maxPages: z3.number().int().positive().max(5e4).optional().describe("Maximum pages crawled and audited. Defaults to 1,000; hard maximum 50,000."),
+  maxPages: siteAuditPageBudgetSchema.optional().describe("Maximum pages crawled and audited (1 to 50,000). Omitted uses the project's saved budget (siteAuditMaxPages, set with canonry_project_upsert or canonry_apply_config), else the full site, up to 50,000; pass a smaller number for a quick look."),
   maxEdges: z3.number().int().positive().max(1e6).optional().describe("Maximum link observations retained for this crawl. When omitted the crawl engine derives the budget from the page count; hard maximum 1,000,000."),
   maxDepth: z3.number().int().min(0).max(100).optional().describe("Maximum internal-link depth from the root page."),
   checkDeadLinks: z3.boolean().optional().describe("Opt in to internal dead-link checks. Omitted and false both disable checks.")
@@ -2694,7 +2695,7 @@ var canonryMcpTools = [
   defineTool({
     name: "canonry_project_upsert",
     title: "Create or replace project",
-    description: "Create or replace a Canonry project. PUT semantics: fields not in the request are reset to their defaults. Provide the full intended project shape. Exceptions: an omitted providerDispatchModes (provider \u2192 sync|batch for scheduled sweeps) keeps the stored preference; send {} to clear it. An omitted qualifiedAliases (the aliases Simple sentiment treats as this brand's own names) keeps the stored list, minus names that no longer qualify (a removed alias, a spelling of the display name, or a competitor's name); send [] to clear it. Competitors are not part of this request, but a new identity (display name, aliases, domains) that claims a tracked competitor's curated alias drops that alias, and the drop is audited.",
+    description: "Create or replace a Canonry project. PUT semantics: fields not in the request are reset to their defaults. Provide the full intended project shape. Exceptions: an omitted providerDispatchModes (provider \u2192 sync|batch for scheduled sweeps) keeps the stored preference; send {} to clear it. An omitted qualifiedAliases (the aliases Simple sentiment treats as this brand's own names) keeps the stored list, minus names that no longer qualify (a removed alias, a spelling of the display name, or a competitor's name); send [] to clear it. Competitors are not part of this request, but a new identity (display name, aliases, domains) that claims a tracked competitor's curated alias drops that alias, and the drop is audited. An omitted siteAuditMaxPages (the Site Health page budget for scans that set none, 1 to 50,000) keeps the stored budget; send null for the full site.",
     access: "write",
     tier: "setup",
     inputSchema: projectUpsertInputSchema,
@@ -2705,7 +2706,7 @@ var canonryMcpTools = [
   defineTool({
     name: "canonry_apply_config",
     title: "Apply project config",
-    description: "Apply one Canonry config-as-code project document. Replaces the project to match the config; fields omitted from the spec are reset to defaults, with these exceptions: a spec with neither queries nor keywords leaves the tracked-query basket unchanged (to clear the basket, pass an explicit empty queries list); an omitted providerDispatchModes keeps the stored preference; an omitted qualifiedAliases keeps the stored list, minus names that no longer qualify (a removed alias, a spelling of the display name, or a competitor's name); send [] to clear it; and a competitor given as a plain domain keeps its stored curated aliases, while `{ domain, aliases }` sets them exactly (`[]` clears). A spec identity that claims a competitor's alias drops that alias (audited); a new competitor whose name overlaps another competitor's alias (one brand key contains the other) or that is the project's own site fails the apply. For multi-document YAML, call this tool once per project document.",
+    description: "Apply one Canonry config-as-code project document. Replaces the project to match the config; fields omitted from the spec are reset to defaults, with these exceptions: a spec with neither queries nor keywords leaves the tracked-query basket unchanged (to clear the basket, pass an explicit empty queries list); an omitted providerDispatchModes keeps the stored preference; an omitted qualifiedAliases keeps the stored list, minus names that no longer qualify (a removed alias, a spelling of the display name, or a competitor's name); send [] to clear it; and a competitor given as a plain domain keeps its stored curated aliases, while `{ domain, aliases }` sets them exactly (`[]` clears). A spec identity that claims a competitor's alias drops that alias (audited); a new competitor whose name overlaps another competitor's alias (one brand key contains the other) or that is the project's own site fails the apply. An omitted siteAuditMaxPages keeps the stored Site Health page budget (null means the full site). For multi-document YAML, call this tool once per project document.",
     access: "write",
     tier: "core",
     inputSchema: applyConfigInputSchema,
@@ -2910,7 +2911,7 @@ var canonryMcpTools = [
   defineTool({
     name: "canonry_measurement_portfolio_summary",
     title: "Summarize measured Properties",
-    description: "Start here for best and worst Properties. One call returns the weakest Properties, the names answers wrote instead of them, and the domains engines cited for them. Defaults to non-brand queries; state the returned queryClass and never pool branded with non-brand. Returns at most 4 Property rows per list, the most that fit one result; a larger limit is lowered and limitNote says so. For other rows pass groupKey (one metro's weakest), or page canonry_measurement_overview. Mention (answer text names the Property) and citation (a source URL on its page) are separate: report each as numerator/denominator, with any unattributed count beside mention. Denominators count answers, one per query per engine (queries x engines = answers). Group Properties only by each row's metro and submarkets, never by label. namedInsteadInAnswerText lists names written in the answer text of answers that neither named nor cited the Property, counted by answer: they are not citations, and namedInsteadInAnswerTextTotal counts distinct names, not answers. citedDomains and weakestAnswerSources count answers citing each domain, every engine included; weakestAnswerSources pools the weakest rows plus every tied Property, each answer once, and its domainTotal counts distinct domains, not citations. weakestAnswerSources.ownDomainAnswers counts those answers that cite any of your own domains. A Property's citationCoverage counts only answers citing that Property's own page, so 0% citation does not mean your domain went uncited; use ownDomainAnswers for that. When tiedAtWeakest is set, tiedAtWeakest.count Properties share the weakest rates and the rows are the first of them by name, not a rank: report the count, tiedAtWeakest.byMetro (tied Properties per metro over the whole tie; a Property in two metros counts in both) and tiedAtWeakest.namedInstead (names written instead across the whole tie, counted by distinct answer; namedInsteadTotal is distinct names), and call the rows examples. mentionRanking.strongest/.weakest rank every Property with an available mention rate; .excluded lists the rest with reasons. markets lists every market at one level, worst-first (every metro, or a groupKey's submarkets), whatever the limit. Markets may share Properties and never sum to portfolio totals. Reads stored data only; never starts provider work.",
+    description: "Start here for best and worst Properties. One call returns the weakest Properties, the names answers wrote instead of them, and the domains engines cited for them. Defaults to non-brand queries; state the returned queryClass and never pool branded with non-brand. Returns at most 4 Property rows per list, the most that fit one result; a larger limit is lowered and limitNote says so. For other rows pass groupKey (one metro's weakest), or page canonry_measurement_overview. Mention (answer text names the Property) and citation (a source URL on its page) are separate: report each as numerator/denominator, with any unattributed count beside mention and any unchecked count beside citation (saved answers whose source capture was incomplete; they are in neither side). Denominators count answers, one per query per engine (queries x engines = answers, less any unattributed or unchecked). Group Properties only by each row's metro and submarkets, never by label. namedInsteadInAnswerText lists names written in the answer text of answers that neither named nor cited the Property, counted by answer: they are not citations, and namedInsteadInAnswerTextTotal counts distinct names, not answers. citedDomains and weakestAnswerSources count answers citing each domain, every engine included; weakestAnswerSources pools the weakest rows plus every tied Property, each answer once, and its domainTotal counts distinct domains, not citations. weakestAnswerSources.ownDomainAnswers counts those answers that cite any of your own domains. A Property's citationCoverage counts only answers citing that Property's own page, so 0% citation does not mean your domain went uncited; use ownDomainAnswers for that. When tiedAtWeakest is set, tiedAtWeakest.count Properties share the weakest rates and the rows are the first of them by name, not a rank: report the count, tiedAtWeakest.byMetro (tied Properties per metro over the whole tie; a Property in two metros counts in both) and tiedAtWeakest.namedInstead (names written instead across the whole tie, counted by distinct answer; namedInsteadTotal is distinct names), and call the rows examples. mentionRanking.strongest/.weakest rank every Property with an available mention rate; .excluded lists the rest with reasons. markets lists every market at one level, worst-first (every metro, or a groupKey's submarkets), whatever the limit. Markets may share Properties and never sum to portfolio totals. Reads stored data only; never starts provider work.",
     access: "read",
     tier: "monitoring",
     inputSchema: measurementPortfolioSummaryInputSchema,
@@ -3815,7 +3816,7 @@ var canonryMcpTools = [
   defineTool({
     name: "canonry_technical_aeo_run",
     title: "Run Technical AEO site audit",
-    description: "Start a site-audit run. The run discovers root, sitemap, and linked pages. It defaults to 1,000 pages, and the engine derives the link-observation budget from that page count with a 100,000 floor, so raising maxPages raises it too. An explicit maxEdges is a ceiling that replaces the derivation and can sit below it. It returns {runId, status} and continues in the background. If an active run has identical effective options, this tool returns it; different options are refused. Poll canonry_run_get, then read the crawl and score tools.",
+    description: "Start a site-audit run. The run discovers root, sitemap, and linked pages. Without maxPages it uses the project's saved page budget (siteAuditMaxPages), else the full site (up to 50,000 pages), and the engine derives the link-observation budget from that page count with a 100,000 floor, so raising maxPages raises it too. An explicit maxEdges is a ceiling that replaces the derivation and can sit below it. It returns {runId, status} and continues in the background. If an active run has identical effective options, this tool returns it; different options are refused. Poll canonry_run_get, then read the crawl and score tools.",
     access: "write",
     tier: "monitoring",
     inputSchema: technicalAeoRunInputSchema,

package/dist/{chunk-QND5LEAC.js → chunk-KS66E5H6.js}

@@ -5,7 +5,7 @@ import {
   siteCrawlGraphNodes,
   siteCrawlPages,
   siteCrawlSnapshots
-} from "./chunk-XODXE6PT.js";
+} from "./chunk-ZHI2MCMD.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-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/site-audit-factors.ts
 function recordedSharePct(factor) {

package/dist/{chunk-MQPBOQUH.js → chunk-OY5QVO34.js}

@@ -20,20 +20,20 @@ import {
   saveConfig,
   saveConfigPatch,
   sentimentInstallReadiness
-} from "./chunk-RVELNJAW.js";
+} from "./chunk-6Z5MRX24.js";
 import {
   computeFactorAverages,
   deleteSiteCrawlGraphLayout,
   persistSiteCrawlGraphLayout,
   prepareSiteCrawlGraphLayout,
   recordedSharePct
-} from "./chunk-QND5LEAC.js";
+} from "./chunk-KS66E5H6.js";
 import {
   CanonryMcpToolNames,
   CliError,
   PACKAGE_VERSION,
   canonryMcpTools
-} from "./chunk-FWV6IIYY.js";
+} from "./chunk-JURAGZOO.js";
 import {
   CC_CACHE_DIR,
   DUCKDB_SPEC,
@@ -205,7 +205,7 @@ import {
   siteLivenessState,
   toAlertView,
   usageCounters
-} from "./chunk-XODXE6PT.js";
+} from "./chunk-ZHI2MCMD.js";
 import {
   AGENT_MEMORY_VALUE_MAX_BYTES,
   AGENT_NONE,
@@ -411,7 +411,7 @@ import {
   visibilityReportRequestSchema,
   winnabilityClassLabel,
   withRetry
-} from "./chunk-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/runtime-context.ts
 function cliRuntimeContext(env = process.env, stdio = process) {
@@ -7869,7 +7869,7 @@ function nonBlank(value) {
   return trimmed ? trimmed : void 0;
 }
 function resolveBuildCommit(env = process.env) {
-  const embedded = true ? "254a7fb5e66f080e53bad263c955da797653b6b0" : void 0;
+  const embedded = true ? "1fae609bb90f155eec013fe7e84214c3306cc9b7" : void 0;
   return nonBlank(embedded) ?? nonBlank(env.CANONRY_COMMIT);
 }
 function resolveInstanceIdentity(env = process.env) {
@@ -14780,7 +14780,7 @@ var Scheduler = class {
           return;
         }
         const runId2 = crypto21.randomUUID();
-        const effectiveRequest = normalizeSiteAuditRunRequest({ maxPages: SITE_AUDIT_MAX_PAGE_LIMIT });
+        const effectiveRequest = normalizeSiteAuditRunRequest({}, project.siteAuditMaxPages);
         this.db.transaction((tx) => {
           tx.insert(runs).values({
             id: runId2,

package/dist/{chunk-XODXE6PT.js → chunk-ZHI2MCMD.js}

@@ -626,6 +626,7 @@ import {
   queryTrackingWorkspaceResponseSchema,
   quotaExceeded,
   rankCriticizedProperties,
+  rateOverChecked,
   recommendationBriefDtoSchema,
   recommendationExplainRequestSchema,
   recommendationExplanationDtoSchema,
@@ -795,7 +796,7 @@ import {
   wordpressSchemaDeployResultDtoSchema,
   wordpressSchemaStatusResultDtoSchema,
   wordpressStatusDtoSchema
-} from "./chunk-YQXQBFHT.js";
+} from "./chunk-3L7QPB37.js";
 
 // src/intelligence-service.ts
 import { eq as eq84, desc as desc37, asc as asc15, and as and71, ne as ne15, or as or20, inArray as inArray33, gte as gte22, lte as lte20, isNull as isNull16, sql as sql31, exists as exists2 } from "drizzle-orm";
@@ -953,6 +954,8 @@ var projects = sqliteTable("projects", {
    * sidecar at dispatch; never read by mention detection or query classes.
    */
   qualifiedAliases: text("qualified_aliases", { mode: "json" }).$type().notNull().default([]),
+  /** Site Health page budget for scans that set none; null means the full site (up to 50,000 pages). */
+  siteAuditMaxPages: integer("site_audit_max_pages"),
   country: text("country").notNull(),
   language: text("language").notNull(),
   tags: text("tags", { mode: "json" }).$type().notNull().default([]),
@@ -8152,6 +8155,16 @@ var MIGRATION_VERSIONS = [
     statements: [
       `ALTER TABLE competitors ADD COLUMN aliases TEXT NOT NULL DEFAULT '[]'`
     ]
+  },
+  {
+    // Saved Site Health page budget for scans that set none. Nullable with no
+    // default: every existing project reads as full site, and an older writer
+    // that never sets it leaves it null.
+    version: 169,
+    name: "projects-site-audit-max-pages",
+    statements: [
+      `ALTER TABLE projects ADD COLUMN site_audit_max_pages INTEGER`
+    ]
   }
 ];
 var AERO_LEDGER_REBUILDS = [
@@ -11041,7 +11054,7 @@ function buildRunHistory(runs2, snapshotsByRunId, limit = DEFAULT_RUN_HISTORY_LI
     }
     const totalCount = queryCited.size;
     const citedCount = [...queryCited.values()].filter(Boolean).length;
-    const citationRate = percentOf(citedCount, totalCount) ?? 0;
+    const citationRate2 = percentOf(citedCount, totalCount) ?? 0;
     const mentionedCount = [...queryMentioned.values()].filter(Boolean).length;
     const mentionRate3 = percentOf(mentionedCount, totalCount) ?? 0;
     return {
@@ -11049,7 +11062,7 @@ function buildRunHistory(runs2, snapshotsByRunId, limit = DEFAULT_RUN_HISTORY_LI
       createdAt: run.createdAt,
       citedCount,
       totalCount,
-      citationRate,
+      citationRate: citationRate2,
       mentionedCount,
       mentionRate: mentionRate3,
       status: run.status
@@ -15719,11 +15732,10 @@ function coverageRate(slots, edges, prepared) {
   if (slots.length === 0) return { numerator: null, denominator: null, rate: null, reason: "no-population" };
   if (!status.complete) return { numerator: null, denominator: null, rate: null, reason: "incomplete" };
   const basis = sourceCompleteSlots(slots, prepared);
-  if (basis.length === 0) return { numerator: null, denominator: null, rate: null, reason: "evidence-incomplete" };
   const edgeIds = new Set(edges.map((edge) => edge.id));
   const assignedSlots = new Set(prepared.evidence.filter((row) => edgeIds.has(row.usageEdgeId) && row.classification === "assigned").map((row) => row.expectedSlotId));
   const numerator = basis.filter((slot) => assignedSlots.has(slot.id)).length;
-  return { numerator, denominator: basis.length, rate: numerator / basis.length };
+  return rateOverChecked(numerator, slots.length, slots.length - basis.length, "unchecked") ?? unavailable("evidence-incomplete");
 }
 function targetCoverageRate(targetIds, slots, edges, prepared) {
   const status = completeness(slots, prepared);
@@ -15751,9 +15763,7 @@ function mentionRate(target, slots, prepared) {
   return attributableMentionRate(numerator, slots.length, unattributed);
 }
 function attributableMentionRate(numerator, answered2, unattributed) {
-  const denominator = answered2 - unattributed;
-  if (denominator <= 0) return unavailable("identity-ambiguous");
-  return unattributed > 0 ? { numerator, denominator, rate: numerator / denominator, unattributed } : { numerator, denominator, rate: numerator / denominator };
+  return rateOverChecked(numerator, answered2, unattributed, "unattributed") ?? unavailable("identity-ambiguous");
 }
 function providersFor(slots) {
   return sortedUnique(slots.map((slot) => slot.provider));
@@ -15867,11 +15877,13 @@ function buildMeasurementOverviewIndexes(input, prepared, onEvidenceIndexed) {
     slots.push(slot);
     slotsByExecutionId.set(slot.executionId, slots);
   }
+  const observedSlotIds = /* @__PURE__ */ new Set();
   const answeredSlotIds = /* @__PURE__ */ new Set();
   const sourceCompleteSlotIds = /* @__PURE__ */ new Set();
   const mentionedSlotIdsByTargetId = /* @__PURE__ */ new Map();
   const unknownMentionSlotIdsByTargetId = /* @__PURE__ */ new Map();
   for (const [slotId, observation] of prepared.observationsBySlot) {
+    observedSlotIds.add(slotId);
     if (observation.sourceComplete) sourceCompleteSlotIds.add(slotId);
     if (observation.input.answerText === null) continue;
     answeredSlotIds.add(slotId);
@@ -15902,6 +15914,7 @@ function buildMeasurementOverviewIndexes(input, prepared, onEvidenceIndexed) {
     targetsById,
     targetEdgesByTargetId,
     slotsByExecutionId,
+    observedSlotIds,
     answeredSlotIds,
     sourceCompleteSlotIds,
     mentionedSlotIdsByTargetId,
@@ -15956,14 +15969,14 @@ function scopeMentionRate(input, targetIds, slots, answered2, prepared, edges) {
 }
 function indexedScopeCitationRate(slots, edges, indexes) {
   if (slots.length === 0) return unavailable("no-population");
+  if (slots.some((slot) => !indexes.observedSlotIds.has(slot.id))) return unavailable("evidence-incomplete");
   const basis = slots.filter((slot) => indexes.sourceCompleteSlotIds.has(slot.id));
-  if (basis.length !== slots.length) return unavailable("evidence-incomplete");
   const assignedSlots = /* @__PURE__ */ new Set();
   for (const edge of edges) {
     for (const slotId of indexes.assignedSlotIdsByEdgeId.get(edge.id) ?? []) assignedSlots.add(slotId);
   }
   const numerator = basis.filter((slot) => assignedSlots.has(slot.id)).length;
-  return { numerator, denominator: basis.length, rate: numerator / basis.length };
+  return rateOverChecked(numerator, slots.length, slots.length - basis.length, "unchecked") ?? unavailable("evidence-incomplete");
 }
 function targetMentionRate(targets, slots, answered2, indexes) {
   const mentionable = targets?.filter((target) => target.aliases.some((alias2) => words(alias2).length > 0)) ?? [];
@@ -17088,6 +17101,7 @@ async function projectRoutes(app, opts) {
         locations: nextLocations,
         defaultLocation: nextDefaultLocation,
         autoExtractBacklinks: body.autoExtractBacklinks ?? false,
+        siteAuditMaxPages: body.siteAuditMaxPages ?? null,
         configSource: body.configSource ?? "api",
         configRevision: 1,
         createdAt: now,
@@ -17160,6 +17174,7 @@ async function projectRoutes(app, opts) {
       });
     }
     const nextAutoExtractBacklinks = body.autoExtractBacklinks !== void 0 ? body.autoExtractBacklinks : existing?.autoExtractBacklinks ?? false;
+    const nextSiteAuditMaxPages = body.siteAuditMaxPages !== void 0 ? body.siteAuditMaxPages : existing?.siteAuditMaxPages ?? null;
     const nextNegativeReviewMaxStars = body.negativeReviewMaxStars !== void 0 ? body.negativeReviewMaxStars : existing ? readNegativeReviewMaxStars(app.db, existing.id) : null;
     const nextMeasurement = body.measurement ?? existing?.measurement ?? DEFAULT_MEASUREMENT_CONFIG;
     const nextAliases = normalizeProjectAliases(body.displayName, body.aliases ?? []);
@@ -17199,6 +17214,7 @@ async function projectRoutes(app, opts) {
           locations: nextLocations,
           defaultLocation: nextDefaultLocation,
           autoExtractBacklinks: nextAutoExtractBacklinks,
+          siteAuditMaxPages: nextSiteAuditMaxPages,
           configSource: body.configSource ?? "api",
           configRevision: existing.configRevision + 1,
           updatedAt: now
@@ -17247,6 +17263,7 @@ async function projectRoutes(app, opts) {
         locations: nextLocations,
         defaultLocation: nextDefaultLocation,
         autoExtractBacklinks: nextAutoExtractBacklinks,
+        siteAuditMaxPages: nextSiteAuditMaxPages,
         configSource: body.configSource ?? "api",
         configRevision: 1,
         createdAt: now,
@@ -17457,6 +17474,7 @@ async function projectRoutes(app, opts) {
         ...project.defaultLocation ? { defaultLocation: project.defaultLocation } : {},
         ...project.autoExtractBacklinks ? { autoExtractBacklinks: true } : {},
         ...negativeReviewMaxStars !== null ? { negativeReviewMaxStars } : {},
+        ...project.siteAuditMaxPages !== null ? { siteAuditMaxPages: project.siteAuditMaxPages } : {},
         notifications: notificationRows.map((row) => {
           const cfg = row.config;
           return {
@@ -17536,6 +17554,7 @@ function formatProject(row, negativeReviewMaxStars) {
     defaultLocation: row.defaultLocation,
     autoExtractBacklinks: row.autoExtractBacklinks,
     negativeReviewMaxStars,
+    siteAuditMaxPages: row.siteAuditMaxPages ?? null,
     configSource: row.configSource,
     configRevision: row.configRevision,
     createdAt: row.createdAt,
@@ -18398,7 +18417,9 @@ function coverageMetric(rate2) {
     numerator: rate2.numerator,
     denominator: rate2.denominator,
     // Answers the mention rate left out because their identity was unresolved.
-    ...rate2.unattributed === void 0 ? {} : { unattributed: rate2.unattributed }
+    ...rate2.unattributed === void 0 ? {} : { unattributed: rate2.unattributed },
+    // Saved answers the citation rate left out because their source capture was incomplete.
+    ...rate2.unchecked === void 0 ? {} : { unchecked: rate2.unchecked }
   };
 }
 function providerRows(property) {
@@ -18562,6 +18583,10 @@ function measurementOutcomeCounts(rows) {
       continue;
     }
     const cited = (citation.numerator ?? citation.value) > 0;
+    if (!cited && citation.unchecked !== void 0) {
+      counts2.notMeasured += 1;
+      continue;
+    }
     if (mentioned && cited) counts2.bothSignals += 1;
     else if (mentioned) counts2.mentionedOnly += 1;
     else if (cited) counts2.citedOnly += 1;
@@ -25478,7 +25503,9 @@ function coverageMetric2(rate2) {
     numerator: rate2.numerator,
     denominator: rate2.denominator,
     // Answers the mention rate left out because their identity was unresolved.
-    ...rate2.unattributed === void 0 ? {} : { unattributed: rate2.unattributed }
+    ...rate2.unattributed === void 0 ? {} : { unattributed: rate2.unattributed },
+    // Saved answers the citation rate left out because their source capture was incomplete.
+    ...rate2.unchecked === void 0 ? {} : { unchecked: rate2.unchecked }
   };
 }
 function countMetric2(rate2) {
@@ -25936,15 +25963,15 @@ function weakestTie(ranked) {
     return { summary: null, rows: [] };
   }
   const mentionRate3 = first.mentionCoverage.value;
-  const citationRate = first.citationCoverage.value;
-  const rows = ranked.filter((row) => row.mentionCoverage.state === "available" && row.mentionCoverage.value === mentionRate3 && row.citationCoverage.state === "available" && row.citationCoverage.value === citationRate);
+  const citationRate2 = first.citationCoverage.value;
+  const rows = ranked.filter((row) => row.mentionCoverage.state === "available" && row.mentionCoverage.value === mentionRate3 && row.citationCoverage.state === "available" && row.citationCoverage.value === citationRate2);
   if (rows.length < 2) return { summary: null, rows: [] };
   const namedInstead = tieNamedInstead(rows);
   return {
     summary: {
       count: rows.length,
       mentionRate: mentionRate3,
-      citationRate,
+      citationRate: citationRate2,
       note: MEASUREMENT_PORTFOLIO_TIE_NOTE,
       byMetro: tieByMetro(rows),
       namedInstead: namedInstead.rows,
@@ -26736,8 +26763,27 @@ function targetValues(candidate, targets, targetKey) {
   }
   return { mention: false, citation };
 }
-function citationForCoverage(candidate, targets, targetKey) {
-  return candidate.observation?.citationComplete ? targetValues(candidate, targets, targetKey).citation : null;
+function citationForCoverage(candidate, targets) {
+  return candidate.observation?.citationComplete ? targetValues(candidate, targets).citation : null;
+}
+function citationRate(candidates, isCited) {
+  if (candidates.length === 0) return unavailable4("no-population");
+  let numerator = 0;
+  let unchecked = 0;
+  for (const candidate of candidates) {
+    if (candidate.observation === null) return unavailable4("evidence-incomplete");
+    if (!candidate.observation.citationComplete) {
+      unchecked++;
+      continue;
+    }
+    const cited = isCited(candidate);
+    if (cited === null) return unavailable4("evidence-incomplete");
+    if (cited) numerator++;
+  }
+  return rateOverChecked(numerator, candidates.length, unchecked, "unchecked") ?? unavailable4("evidence-incomplete");
+}
+function targetCitationRate(candidates, targets, targetKey) {
+  return citationRate(candidates, (candidate) => targetValues(candidate, targets, targetKey).citation);
 }
 function mentionRate2(candidates, targets, targetKey) {
   const values = candidates.map((candidate) => targetValues(candidate, targets, targetKey));
@@ -26751,9 +26797,7 @@ function mentionRate2(candidates, targets, targetKey) {
       unattributed++;
     }
   }
-  const denominator = values.length - unattributed;
-  if (denominator === 0) return unavailable4("identity-ambiguous");
-  return unattributed > 0 ? { numerator, denominator, rate: numerator / denominator, unattributed } : { numerator, denominator, rate: numerator / denominator };
+  return rateOverChecked(numerator, values.length, unattributed, "unattributed") ?? unavailable4("identity-ambiguous");
 }
 function answered(candidate) {
   return candidate.observation !== null;
@@ -26768,16 +26812,18 @@ function targetMetrics(candidates, targets, targetKey) {
   const own = candidates.filter((candidate) => candidate.edges.some((edge) => edge.targetKey === targetKey));
   return {
     mentionCoverage: mentionRate2(own, targets, targetKey),
-    citationCoverage: rate(own.map((candidate) => citationForCoverage(candidate, targets, targetKey)), own.length)
+    citationCoverage: targetCitationRate(own, targets, targetKey)
   };
 }
 function targetPresence(candidates, targets, targetKey) {
-  const values = candidates.filter((candidate) => candidate.edges.some((edge) => edge.targetKey === targetKey)).map((candidate) => targetValues(candidate, targets, targetKey));
+  const scoped = candidates.filter((candidate) => candidate.edges.some((edge) => edge.targetKey === targetKey));
+  const values = scoped.map((candidate) => targetValues(candidate, targets, targetKey));
+  const citations = scoped.map((candidate, index2) => candidate.observation?.citationComplete === false ? null : values[index2].citation);
   const anyKnown = (signals) => signals.includes(true) ? true : signals.length === 0 || signals.includes(null) ? null : false;
   return {
     targetKey,
     mention: anyKnown(values.map((value) => value.mention)),
-    citation: anyKnown(values.map((value) => value.citation)),
+    citation: anyKnown(citations),
     identityAmbiguous: values.some((value) => value.mentionUnavailableReason === "identity-ambiguous")
   };
 }
@@ -26806,7 +26852,7 @@ function coverageSummary(candidates, definition, targets = targetMap(definition)
     queryCount: new Set(rows.map((candidate) => candidate.slot.queryKey)).size,
     answerCount: rows.filter(answered).length,
     mentionCoverage: mentionRate2(rows, targets),
-    citationCoverage: rate(rows.map((candidate) => citationForCoverage(candidate, targets)), rows.length)
+    citationCoverage: targetCitationRate(rows, targets)
   };
 }
 function summary2(candidates, definition, populationTargetKeys = selectedTargetKeys(candidates)) {
@@ -26902,15 +26948,12 @@ function competitorRows(candidates) {
       const observation = candidate.observation;
       return observation === null || observation.answerText === null ? null : observation.competitorMentionDomains.includes(domain);
     });
-    const citationValues = unique3.map((candidate) => {
-      const observation = candidate.observation;
-      return observation === null || !observation.citationComplete ? null : observation.competitorCitationDomains.includes(domain);
-    });
     return {
       domain,
       answerCount: unique3.filter(answered).length,
       mentionCoverage: rate(mentionValues, unique3.length),
-      citationCoverage: rate(citationValues, unique3.length)
+      // The Property rule: an unchecked answer leaves both sides of the rate.
+      citationCoverage: citationRate(unique3, (candidate) => candidate.observation.competitorCitationDomains.includes(domain))
     };
   });
 }
@@ -26950,7 +26993,7 @@ function breakdown(candidates, definition, populationTargetKeys = selectedTarget
     const groupTargetKeys = new Set(targetKeys);
     const own = candidates.map((candidate) => narrowedToTargetKeys(candidate, groupTargetKeys)).filter((candidate) => candidate !== null);
     const metrics = mentionRate2(own, targets);
-    const citations = rate(own.map((candidate) => citationForCoverage(candidate, targets)), own.length);
+    const citations = targetCitationRate(own, targets);
     return {
       id: group.id,
       label: group.label,
@@ -27578,6 +27621,8 @@ function frozenSimpleRun(run, definition, snapshots, includeEvidence = true) {
       mentionComplete: snapshot.answerText !== null || snapshot.answerMentioned !== null,
       mentionedTargetKeys: mentioned ? ["project"] : [],
       citedTargetKeys: snapshot.citationState === "cited" ? ["project"] : [],
+      // Simple citation state comes from the stored cited domains, not from
+      // source-URL capture, so a saved Simple answer is never `unchecked`.
       citationComplete: true,
       competitorMentionDomains: competitorSignalsForSnapshot.mentioned,
       competitorCitationDomains: competitorSignalsForSnapshot.cited,
@@ -27660,6 +27705,7 @@ function legacySimpleRun(project, run, snapshots, includeEvidence = true) {
       mentionComplete: snapshot.answerMentioned !== null,
       mentionedTargetKeys: snapshot.answerMentioned === true ? ["project"] : [],
       citedTargetKeys: snapshot.citationState === "cited" ? ["project"] : [],
+      // Legacy rows carry only the stored citation state (see the frozen-simple note).
       citationComplete: true,
       competitorMentionDomains: [],
       competitorCitationDomains: [],
@@ -29977,6 +30023,8 @@ async function applyRoutes(app, opts) {
           locations: config.spec.locations ?? [],
           defaultLocation: config.spec.defaultLocation ?? null,
           autoExtractBacklinks: config.spec.autoExtractBacklinks ?? false,
+          // Present sets (null = full site); absent leaves the stored budget alone.
+          ...config.spec.siteAuditMaxPages !== void 0 ? { siteAuditMaxPages: config.spec.siteAuditMaxPages } : {},
           configSource: "config-file",
           configRevision: existing.configRevision + 1,
           updatedAt: now
@@ -30010,6 +30058,7 @@ async function applyRoutes(app, opts) {
           locations: config.spec.locations ?? [],
           defaultLocation: config.spec.defaultLocation ?? null,
           autoExtractBacklinks: config.spec.autoExtractBacklinks ?? false,
+          siteAuditMaxPages: config.spec.siteAuditMaxPages ?? null,
           configSource: "config-file",
           configRevision: 1,
           createdAt: now,
@@ -30156,6 +30205,7 @@ async function applyRoutes(app, opts) {
       defaultLocation: project.defaultLocation,
       autoExtractBacklinks: project.autoExtractBacklinks,
       negativeReviewMaxStars: readNegativeReviewMaxStars(app.db, project.id),
+      siteAuditMaxPages: project.siteAuditMaxPages ?? null,
       configSource: project.configSource,
       configRevision: project.configRevision,
       createdAt: project.createdAt,
@@ -35327,6 +35377,7 @@ function formatProject2(row, negativeReviewMaxStars) {
     defaultLocation: row.defaultLocation,
     autoExtractBacklinks: row.autoExtractBacklinks,
     negativeReviewMaxStars,
+    siteAuditMaxPages: row.siteAuditMaxPages ?? null,
     configSource: row.configSource,
     configRevision: row.configRevision,
     createdAt: row.createdAt,
@@ -43838,7 +43889,7 @@ var routeCatalog = [
             properties: {
               sitemapUrl: { ...stringSchema, description: "Deprecated compatibility alias for the sitemap override." },
               limit: { ...integerSchema, minimum: 1, maximum: 2e3, description: "Deprecated compatibility alias for maxPages. Max 2000." },
-              maxPages: { ...integerSchema, minimum: 1, maximum: 5e4, description: "Crawl page budget. Defaults to 1000; max 50000." },
+              maxPages: { ...integerSchema, minimum: 1, maximum: 5e4, description: "Crawl page budget. When omitted, the project's saved budget (siteAuditMaxPages) applies, else the full site; max 50000." },
               maxEdges: { ...integerSchema, minimum: 1, maximum: 1e6, description: "Internal-link observation budget. When omitted the crawl engine derives it from the page count; max 1000000." },
               maxDepth: { ...integerSchema, minimum: 0, maximum: 100, description: "Maximum crawl depth from the root. Max 100." },
               checkDeadLinks: { ...booleanSchema, default: false, description: "Enable dead-link checking. Defaults to false." }
@@ -86752,7 +86803,7 @@ async function technicalAeoRoutes(app, opts) {
         reason: "no-site-audit-handler"
       });
     }
-    const effectiveRequest = normalizeSiteAuditRunRequest(parsed.data);
+    const effectiveRequest = normalizeSiteAuditRunRequest(parsed.data, project.siteAuditMaxPages);
     const identityKey = siteAuditRequestIdentity(effectiveRequest);
     const result = app.db.transaction((tx) => {
       const existing = tx.select({