@pas7/llm-seo 0.1.6 → 0.3.0

package/dist/cli/bin.js

@@ -48,9 +48,101 @@ var init_exit_codes = __esm({
     };
   }
 });
-var SiteConfigSchema, BrandConfigSchema, SectionsConfigSchema, SocialConfigSchema, ContactConfigSchema, RestrictedClaimsConfigSchema, PolicyConfigSchema, BookingConfigSchema, MachineHintsConfigSchema, OutputPathsConfigSchema, OutputConfigSchema, FormatConfigSchema, LlmsSeoConfigSchema, ConfigSchema, LocaleConfigSchema, CheckConfigSchema;
+var ISO_DATE_OR_DATETIME, ManifestItemSchema, ManifestSourceSchema, ManifestValueSchema, OptionalSectionSchema, PageManifestSchema;
+var init_manifest_schema = __esm({
+  "src/schema/manifest.schema.ts"() {
+    ISO_DATE_OR_DATETIME = /^\d{4}-\d{2}-\d{2}(?:T\d{2}:\d{2}(?::\d{2}(?:\.\d{1,3})?)?(?:Z|[+-]\d{2}:\d{2})?)?$/;
+    ManifestItemSchema = z.object({
+      /** URL path slug - required */
+      slug: z.string().min(1, { message: "Slug is required and cannot be empty" }),
+      /** Available locales for this page */
+      locales: z.array(z.string().min(2)).optional(),
+      /** Publication date (ISO 8601) */
+      publishedAt: z.string().regex(
+        ISO_DATE_OR_DATETIME,
+        { message: "publishedAt must be ISO date (YYYY-MM-DD) or ISO datetime" }
+      ).optional(),
+      /** Last update date (ISO 8601) */
+      updatedAt: z.string().regex(
+        ISO_DATE_OR_DATETIME,
+        { message: "updatedAt must be ISO date (YYYY-MM-DD) or ISO datetime" }
+      ).optional(),
+      /** Override canonical URL */
+      canonicalOverride: z.string().url().optional(),
+      /** Priority for citations (0-100) */
+      priority: z.number().int().min(0).max(100).optional(),
+      /** Title for display */
+      title: z.string().optional(),
+      /** Description for display */
+      description: z.string().optional()
+    });
+    ManifestSourceSchema = z.object({
+      /** Source type */
+      type: z.enum(["file", "url", "module"]),
+      /** Source location */
+      source: z.string(),
+      /** Optional transform function */
+      transform: z.string().optional()
+    });
+    ManifestValueSchema = z.union([
+      z.array(ManifestItemSchema),
+      ManifestSourceSchema
+    ]);
+    z.record(z.string(), ManifestValueSchema);
+    OptionalSectionSchema = z.object({
+      /** Section title */
+      title: z.string().min(1),
+      /** Section content */
+      content: z.string().optional()
+    });
+    PageManifestSchema = z.object({
+      /** Page path (e.g., /about) */
+      path: z.string().startsWith("/"),
+      /** Page title */
+      title: z.string().optional(),
+      /** Page description */
+      description: z.string().optional(),
+      /** Page content for llms-full.txt */
+      content: z.string().optional(),
+      /** Whether page is optional (included only in full) */
+      optional: z.boolean().default(false),
+      /** Last modified timestamp (ISO 8601) */
+      lastModified: z.string().datetime().optional()
+    });
+    z.object({
+      /** Site base URL */
+      baseUrl: z.string().url(),
+      /** Site title */
+      title: z.string().min(1),
+      /** Site description */
+      description: z.string().optional(),
+      /** List of pages */
+      pages: z.array(PageManifestSchema).min(1),
+      /** Optional sections for llms.txt */
+      optionalSections: z.array(OptionalSectionSchema).optional(),
+      /** Site version */
+      version: z.string().optional(),
+      /** Generation timestamp (ISO 8601) */
+      generatedAt: z.string().datetime().optional()
+    });
+    z.object({
+      /** Build ID */
+      buildId: z.string(),
+      /** List of static pages */
+      pages: z.record(z.string(), z.array(z.string())),
+      /** Dynamic routes */
+      dynamicRoutes: z.record(z.string(), z.object({
+        routeRegex: z.string(),
+        dataRoute: z.string(),
+        dataRouteRegex: z.string()
+      })).optional()
+    });
+  }
+});
+var SiteConfigSchema, BrandConfigSchema, SectionsConfigSchema, RouteStyleSchema, ManifestSectionConfigSchema, ManifestConfigValueSchema, ManifestsConfigSchema2, SocialConfigSchema, ContactConfigSchema, RestrictedClaimsConfigSchema, PolicyConfigSchema, BookingConfigSchema, MachineHintsConfigSchema, OutputPathsConfigSchema, OutputConfigSchema, FormatConfigSchema, LlmsSeoConfigSchema, ConfigSchema, LocaleConfigSchema, CheckConfigSchema;
 var init_config_schema = __esm({
   "src/schema/config.schema.ts"() {
+    init_manifest_schema();
     SiteConfigSchema = z.object({
       /** Site base URL - must be valid URL with http/https, no trailing slash */
       baseUrl: z.string().url({ message: "Must be a valid URL with http or https protocol" }).refine(
@@ -76,6 +168,29 @@ var init_config_schema = __esm({
       /** Hub paths - e.g., ["/services", "/blog", "/projects"] */
       hubs: z.array(z.string()).default([])
     });
+    RouteStyleSchema = z.enum(["prefix", "suffix", "locale-segment", "custom"]);
+    ManifestSectionConfigSchema = z.object({
+      /** Items in this manifest section */
+      items: z.array(ManifestItemSchema),
+      /** Optional display name for section */
+      sectionName: z.string().optional(),
+      /** Routing behavior for this section */
+      routeStyle: RouteStyleSchema.optional(),
+      /** Section path prefix, e.g. "/blog" */
+      sectionPath: z.string().optional(),
+      /** Per-section default locale override */
+      defaultLocaleOverride: z.string().min(2).optional(),
+      /** Custom pathname factory for routeStyle="custom" */
+      pathnameFor: z.custom(
+        (value) => value === void 0 || typeof value === "function",
+        { message: "pathnameFor must be a function" }
+      ).optional()
+    });
+    ManifestConfigValueSchema = z.union([
+      z.array(ManifestItemSchema),
+      ManifestSectionConfigSchema
+    ]);
+    ManifestsConfigSchema2 = z.record(z.string(), ManifestConfigValueSchema);
     SocialConfigSchema = z.object({
       /** Twitter handle or URL */
       twitter: z.string().optional(),
@@ -152,7 +267,7 @@ var init_config_schema = __esm({
       /** Sections configuration */
       sections: SectionsConfigSchema.optional(),
       /** Manifests configuration */
-      manifests: z.record(z.unknown()).default({}),
+      manifests: ManifestsConfigSchema2.default({}),
       /** Contact configuration */
       contact: ContactConfigSchema.optional(),
       /** Policy configuration */
@@ -629,18 +744,6 @@ var init_locale = __esm({
 function dedupeUrls(urls) {
   return [...new Set(urls)];
 }
-function buildLocalePrefix(locale, strategy, defaultLocale) {
-  if (strategy === "none") {
-    return "";
-  }
-  if (strategy === "subdomain") {
-    return "";
-  }
-  if (strategy === "prefix" && locale === defaultLocale) {
-    return "";
-  }
-  return `/${locale}`;
-}
 function buildBaseUrlWithSubdomain(baseUrl, locale, strategy, defaultLocale) {
   if (strategy !== "subdomain" || locale === defaultLocale) {
     return baseUrl;
@@ -653,7 +756,17 @@ function buildBaseUrlWithSubdomain(baseUrl, locale, strategy, defaultLocale) {
   }
 }
 function createCanonicalUrlForItem(item, options) {
-  const { baseUrl, routePrefix, defaultLocale, trailingSlash, localeStrategy } = options;
+  const {
+    baseUrl,
+    routePrefix,
+    defaultLocale,
+    trailingSlash,
+    localeStrategy,
+    routeStyle,
+    sectionName,
+    sectionPath,
+    pathnameFor
+  } = options;
   if (item.canonicalOverride && typeof item.canonicalOverride === "string") {
     return item.canonicalOverride;
   }
@@ -663,33 +776,79 @@ function createCanonicalUrlForItem(item, options) {
     availableLocales
   });
   const locale = canonicalLocale ?? defaultLocale;
-  const localePrefix = buildLocalePrefix(locale, localeStrategy, defaultLocale);
+  const sectionBase = normalizeSectionPath(sectionPath ?? routePrefix ?? "");
+  const effectiveRouteStyle = routeStyle ?? inferRouteStyleFromLocaleStrategy(localeStrategy);
+  const normalizedSlug = normalizeItemSlug(item.slug, sectionBase);
+  const customPath = effectiveRouteStyle === "custom" ? pathnameFor?.({
+    item,
+    sectionName: sectionName ?? "",
+    slug: normalizedSlug,
+    locale,
+    defaultLocale,
+    sectionPath: sectionBase
+  }) : void 0;
+  const resolvedPath = customPath ?? buildPathFromRouteStyle({
+    routeStyle: effectiveRouteStyle,
+    sectionPath: sectionBase,
+    slug: normalizedSlug,
+    locale,
+    defaultLocale
+  });
   const effectiveBaseUrl = buildBaseUrlWithSubdomain(baseUrl, locale, localeStrategy, defaultLocale);
-  const parts = [];
-  if (localePrefix) {
-    parts.push(localePrefix);
-  }
-  if (routePrefix) {
-    parts.push(routePrefix);
-  }
-  parts.push(item.slug);
-  const fullPath = joinUrlParts(...parts);
   return normalizeUrl({
     baseUrl: effectiveBaseUrl,
-    path: fullPath,
+    path: resolvedPath,
     trailingSlash,
     stripQuery: true,
     stripHash: true
   });
 }
-function createCanonicalUrlsFromManifest(options) {
-  const { items } = options;
-  if (!items || items.length === 0) {
-    return [];
+function inferRouteStyleFromLocaleStrategy(strategy) {
+  if (strategy !== "prefix") {
+    return "custom";
+  }
+  return "prefix";
+}
+function buildPathFromRouteStyle(args) {
+  const { routeStyle, sectionPath, slug, locale, defaultLocale } = args;
+  const includeLocale = locale !== defaultLocale;
+  if (routeStyle === "locale-segment") {
+    return joinUrlParts(sectionPath, locale, slug);
+  }
+  if (routeStyle === "suffix") {
+    if (includeLocale) {
+      return joinUrlParts(sectionPath, slug, locale);
+    }
+    return joinUrlParts(sectionPath, slug);
+  }
+  if (routeStyle === "custom") {
+    return joinUrlParts(sectionPath, slug);
+  }
+  if (includeLocale) {
+    return joinUrlParts(locale, sectionPath, slug);
+  }
+  return joinUrlParts(sectionPath, slug);
+}
+function normalizeSectionPath(sectionPath) {
+  if (!sectionPath || sectionPath === "/") {
+    return "";
+  }
+  return sectionPath.startsWith("/") ? sectionPath : `/${sectionPath}`;
+}
+function normalizeItemSlug(slug, sectionPath) {
+  const normalizedSlug = slug.startsWith("/") ? slug : `/${slug}`;
+  if (!sectionPath) {
+    return normalizedSlug;
+  }
+  if (normalizedSlug === sectionPath) {
+    return "/";
+  }
+  const withSlash = `${sectionPath}/`;
+  if (normalizedSlug.startsWith(withSlash)) {
+    const relative = normalizedSlug.slice(withSlash.length);
+    return relative ? `/${relative}` : "/";
   }
-  const urls = items.map((item) => createCanonicalUrlForItem(item, options));
-  const deduped = dedupeUrls(urls);
-  return sortUrlsByPath(deduped);
+  return normalizedSlug;
 }
 var init_canonical_from_manifest = __esm({
   "src/core/canonical/canonical-from-manifest.ts"() {
@@ -699,6 +858,96 @@ var init_canonical_from_manifest = __esm({
   }
 });
 
+// src/core/canonical/config-manifests.ts
+function resolveManifestSections(config) {
+  const sections = [];
+  const manifestEntries = Object.entries(config.manifests).sort(
+    (a, b) => a[0].localeCompare(b[0], "en", { sensitivity: "base", numeric: true })
+  );
+  for (const [key, rawValue] of manifestEntries) {
+    const value = rawValue;
+    const resolved = resolveOneManifestSection(key, value);
+    sections.push(resolved);
+  }
+  return sections;
+}
+function createCanonicalBundleFromConfig(config) {
+  const sections = resolveManifestSections(config);
+  const entries = [];
+  const defaultLocale = config.site.defaultLocale ?? config.brand.locales[0] ?? "en";
+  const trailingSlash = config.format?.trailingSlash ?? "never";
+  const localeStrategy = config.format?.localeStrategy ?? "prefix";
+  for (const section of sections) {
+    const sectionDefaultLocale = section.defaultLocaleOverride ?? defaultLocale;
+    const sortedItems = sortBy(section.items, (item) => {
+      const localesKey = (item.locales ?? []).join(",");
+      return `${item.slug}|${localesKey}`;
+    });
+    for (const item of sortedItems) {
+      const itemOptions = {
+        baseUrl: config.site.baseUrl,
+        sectionName: section.sectionName,
+        defaultLocale: sectionDefaultLocale,
+        trailingSlash,
+        localeStrategy,
+        ...section.sectionPath && {
+          routePrefix: section.sectionPath,
+          sectionPath: section.sectionPath
+        },
+        ...section.routeStyle && { routeStyle: section.routeStyle },
+        ...section.pathnameFor && { pathnameFor: section.pathnameFor }
+      };
+      const canonicalUrl = createCanonicalUrlForItem(item, itemOptions);
+      entries.push({
+        sectionKey: section.key,
+        sectionName: section.sectionName,
+        item,
+        canonicalUrl
+      });
+    }
+  }
+  const sortedEntries = sortBy(entries, (entry) => {
+    return `${entry.canonicalUrl}|${entry.sectionKey}|${entry.item.slug}`;
+  });
+  const canonicalUrls = sortUrlsByPath(dedupeUrls(sortedEntries.map((entry) => entry.canonicalUrl)));
+  const manifestItems = sortedEntries.map((entry) => entry.item);
+  return {
+    canonicalUrls,
+    manifestItems,
+    entries: sortedEntries
+  };
+}
+function resolveOneManifestSection(key, value) {
+  if (Array.isArray(value)) {
+    return {
+      key,
+      sectionName: key,
+      items: normalizeManifestItems(value)
+    };
+  }
+  return {
+    key,
+    sectionName: value.sectionName ?? key,
+    items: normalizeManifestItems(value.items),
+    ...value.sectionPath && { sectionPath: value.sectionPath },
+    ...value.routeStyle && { routeStyle: value.routeStyle },
+    ...value.defaultLocaleOverride && { defaultLocaleOverride: value.defaultLocaleOverride },
+    ...value.pathnameFor && { pathnameFor: value.pathnameFor }
+  };
+}
+function normalizeManifestItems(items) {
+  return items.map((item) => {
+    const normalizedSlug = item.slug.startsWith("/") ? item.slug : `/${item.slug}`;
+    return { ...item, slug: normalizedSlug };
+  });
+}
+var init_config_manifests = __esm({
+  "src/core/canonical/config-manifests.ts"() {
+    init_canonical_from_manifest();
+    init_sort();
+  }
+});
+
 // src/core/generate/llms-txt.ts
 function createLlmsTxt(options) {
   const { config, canonicalUrls } = options;
@@ -839,7 +1088,7 @@ var init_llms_txt = __esm({
 
 // src/core/generate/llms-full-txt.ts
 function createLlmsFullTxt(options) {
-  const { config, canonicalUrls, manifestItems } = options;
+  const { config, canonicalUrls, manifestItems, manifestEntries } = options;
   const lineEndings = config.format?.lineEndings ?? "lf";
   const lines = [];
   lines.push(`# ${config.brand.name} - Full LLM Context`);
@@ -856,6 +1105,10 @@ function createLlmsFullTxt(options) {
     lines.push(`Organization: ${config.brand.org}`);
   }
   lines.push(`Locales: ${config.brand.locales.join(", ")}`);
+  const lastUpdated = getLastUpdatedDate(manifestItems);
+  if (lastUpdated) {
+    lines.push(`Last Updated: ${lastUpdated}`);
+  }
   lines.push("");
   if (canonicalUrls.length > 0) {
     lines.push("## All Canonical URLs");
@@ -893,11 +1146,16 @@ function createLlmsFullTxt(options) {
       lines.push("");
     }
   }
-  const hasSocial = config.contact?.social?.twitter || config.contact?.social?.linkedin || config.contact?.social?.github;
-  const hasBooking = config.booking?.url;
-  if (hasSocial || hasBooking) {
-    lines.push("## Social & Booking");
+  const hasContact = config.contact?.email || config.contact?.phone || config.contact?.social?.twitter || config.contact?.social?.linkedin || config.contact?.social?.github || config.booking?.url;
+  if (hasContact) {
+    lines.push("## Contact");
     lines.push("");
+    if (config.contact?.email) {
+      lines.push(`- Email: ${config.contact.email}`);
+    }
+    if (config.contact?.phone) {
+      lines.push(`- Phone: ${config.contact.phone}`);
+    }
     if (config.contact?.social?.twitter) {
       lines.push(`- Twitter: ${config.contact.social.twitter}`);
     }
@@ -941,7 +1199,16 @@ function createLlmsFullTxt(options) {
         lines.push(`- [${hub}](${hub}) - ${getHubLabel2(hub)}`);
       }
     }
-    if (manifestItems.length > 0) {
+    if (manifestEntries && manifestEntries.length > 0) {
+      const sortedEntries = sortBy(manifestEntries, (entry) => {
+        return `${entry.item.slug}|${entry.canonicalUrl}`;
+      });
+      for (const entry of sortedEntries) {
+        const title = entry.item.title ?? entry.item.slug;
+        const locales = entry.item.locales?.join(", ") ?? config.brand.locales[0] ?? "en";
+        lines.push(`- [${title}](${entry.canonicalUrl}) (${locales})`);
+      }
+    } else if (manifestItems.length > 0) {
       const sortedItems = sortBy(manifestItems, (item) => item.slug);
       for (const item of sortedItems) {
         const url = item.canonicalOverride ?? `${config.site.baseUrl}${item.slug}`;
@@ -962,6 +1229,24 @@ function createLlmsFullTxt(options) {
     lineCount: finalLines.length
   };
 }
+function getLastUpdatedDate(items) {
+  const timestamps = [];
+  for (const item of items) {
+    if (item.updatedAt) {
+      timestamps.push(item.updatedAt);
+    } else if (item.publishedAt) {
+      timestamps.push(item.publishedAt);
+    }
+  }
+  if (timestamps.length === 0) {
+    return null;
+  }
+  const latest = timestamps.map((value) => new Date(value)).filter((value) => !Number.isNaN(value.getTime())).sort((a, b) => b.getTime() - a.getTime())[0];
+  if (!latest) {
+    return null;
+  }
+  return latest.toISOString().slice(0, 10);
+}
 function getHubLabel2(hub) {
   const labels = {
     "/services": "Services overview",
@@ -995,14 +1280,20 @@ var init_llms_full_txt = __esm({
 
 // src/core/generate/citations.ts
 function createCitationsJson(options) {
-  const { config, manifestItems, sectionName, fixedTimestamp } = options;
-  const sources = manifestItems.map((item) => {
-    const url = item.canonicalOverride ?? `${config.site.baseUrl}${item.slug}`;
+  const { config, manifestItems, sectionName, fixedTimestamp, entries } = options;
+  const sourceInput = entries ?? manifestItems.map((item) => ({
+    item,
+    canonicalUrl: item.canonicalOverride ?? `${config.site.baseUrl}${item.slug}`,
+    sectionName
+  }));
+  const sources = sourceInput.map((entry) => {
+    const url = entry.canonicalUrl;
+    const item = entry.item;
     const defaultLocale = config.site.defaultLocale ?? config.brand.locales[0] ?? "en";
     return {
       url,
       priority: item.priority ?? 50,
-      section: sectionName,
+      section: entry.sectionName,
       locale: item.locales?.[0] ?? defaultLocale,
       ...item.publishedAt && { publishedAt: item.publishedAt },
       ...item.updatedAt && { updatedAt: item.updatedAt },
@@ -1593,6 +1884,7 @@ var init_core = __esm({
     init_sort();
     init_text();
     init_canonical_from_manifest();
+    init_config_manifests();
     init_locale();
     init_llms_txt();
     init_llms_full_txt();
@@ -1636,32 +1928,27 @@ async function generateCommand(options) {
       printVerbose(`Site: ${config.site.baseUrl}`);
       printVerbose(`Brand: ${config.brand.name}`);
     }
-    const manifestItems = extractManifestItems(config);
+    const canonicalBundle = createCanonicalBundleFromConfig(config);
+    const manifestItems = canonicalBundle.manifestItems;
+    const canonicalUrls = canonicalBundle.canonicalUrls;
     if (verbose) {
       printVerbose(`Found ${manifestItems.length} manifest items`);
-    }
-    const canonicalUrls = createCanonicalUrlsFromManifest({
-      items: manifestItems,
-      baseUrl: config.site.baseUrl,
-      defaultLocale: config.site.defaultLocale ?? config.brand.locales[0] ?? "en",
-      trailingSlash: config.format?.trailingSlash ?? "never",
-      localeStrategy: "prefix"
-    });
-    if (verbose) {
       printVerbose(`Generated ${canonicalUrls.length} canonical URLs`);
     }
     const llmsTxtResult = createLlmsTxt({ config, canonicalUrls });
     const llmsFullTxtResult = createLlmsFullTxt({
       config,
       canonicalUrls,
-      manifestItems
+      manifestItems,
+      manifestEntries: canonicalBundle.entries
     });
     let citationsContent = null;
     if (emitCitations) {
       citationsContent = createCitationsJsonString({
         config,
         manifestItems,
-        sectionName: "all"
+        sectionName: "all",
+        entries: canonicalBundle.entries
       });
     }
     const result = {
@@ -1724,70 +2011,6 @@ async function generateCommand(options) {
     return ExitCodes.GENERATION_FAILED;
   }
 }
-function extractManifestItems(config) {
-  const items = [];
-  const manifests = config.manifests;
-  for (const [_manifestName, manifestData] of Object.entries(manifests)) {
-    if (typeof manifestData === "object" && manifestData !== null) {
-      const data = manifestData;
-      if (Array.isArray(data.pages)) {
-        for (const page of data.pages) {
-          const normalized = toManifestItem(page);
-          if (normalized) {
-            items.push(normalized);
-          }
-        }
-      }
-      if (Array.isArray(data)) {
-        for (const item of data) {
-          const normalized = toManifestItem(item);
-          if (normalized) {
-            items.push(normalized);
-          }
-        }
-      }
-    }
-  }
-  return items;
-}
-function toManifestItem(value) {
-  if (typeof value !== "object" || value === null) {
-    return null;
-  }
-  const data = value;
-  const rawSlug = data.slug ?? data.path;
-  if (typeof rawSlug !== "string" || rawSlug.length === 0) {
-    return null;
-  }
-  const item = {
-    slug: rawSlug.startsWith("/") ? rawSlug : `/${rawSlug}`
-  };
-  if (typeof data.title === "string") {
-    item.title = data.title;
-  }
-  if (typeof data.description === "string") {
-    item.description = data.description;
-  }
-  if (Array.isArray(data.locales)) {
-    const locales = data.locales.filter((loc) => typeof loc === "string");
-    if (locales.length > 0) {
-      item.locales = locales;
-    }
-  }
-  if (typeof data.canonicalOverride === "string") {
-    item.canonicalOverride = data.canonicalOverride;
-  }
-  if (typeof data.publishedAt === "string") {
-    item.publishedAt = data.publishedAt;
-  }
-  if (typeof data.updatedAt === "string") {
-    item.updatedAt = data.updatedAt;
-  }
-  if (typeof data.priority === "number") {
-    item.priority = data.priority;
-  }
-  return item;
-}
 var init_generate = __esm({
   "src/cli/commands/generate.ts"() {
     init_load_config();
@@ -1804,7 +2027,7 @@ __export(check_exports, {
   checkCommand: () => checkCommand
 });
 async function checkCommand(options) {
-  const { config: configPath, failOn, verbose } = options;
+  const { config: configPath, failOn, checkMachineHintsLive, verbose } = options;
   try {
     if (verbose) {
       printVerbose(`Loading config from: ${configPath}`);
@@ -1841,19 +2064,15 @@ async function checkCommand(options) {
       ...config.output.paths.citations && { citationsPath: config.output.paths.citations }
     };
     const result = await checkGeneratedFiles(checkOptions);
-    const manifestItems = extractManifestItems2(config);
-    const canonicalUrls = createCanonicalUrlsFromManifest({
-      items: manifestItems,
-      baseUrl: config.site.baseUrl,
-      defaultLocale: config.site.defaultLocale ?? config.brand.locales[0] ?? "en",
-      trailingSlash: config.format?.trailingSlash ?? "never",
-      localeStrategy: "prefix"
-    });
+    const canonicalBundle = createCanonicalBundleFromConfig(config);
+    const manifestItems = canonicalBundle.manifestItems;
+    const canonicalUrls = canonicalBundle.canonicalUrls;
     const expectedLlms = createLlmsTxt({ config, canonicalUrls });
     const expectedLlmsFull = createLlmsFullTxt({
       config,
       canonicalUrls,
-      manifestItems
+      manifestItems,
+      manifestEntries: canonicalBundle.entries
     });
     const requiredMissing = result.issues.some((issue) => {
       return issue.code === "file_missing" && (issue.path === config.output.paths.llmsTxt || issue.path === config.output.paths.llmsFullTxt);
@@ -1883,6 +2102,23 @@ async function checkCommand(options) {
         };
       }
     }
+    if (checkMachineHintsLive) {
+      const liveIssues = await checkMachineHintsLiveEndpoints(config, verbose);
+      if (liveIssues.length > 0) {
+        const issues = [...merged.issues, ...liveIssues];
+        const counts = countSeverities(issues);
+        merged = {
+          ...merged,
+          issues,
+          summary: {
+            ...merged.summary,
+            errors: counts.error,
+            warnings: counts.warning,
+            info: counts.info
+          }
+        };
+      }
+    }
     printCheckReport(merged, verbose);
     if (merged.summary.errors > 0) {
       return ExitCodes.ERROR;
@@ -1897,69 +2133,43 @@ async function checkCommand(options) {
     return ExitCodes.ERROR;
   }
 }
-function extractManifestItems2(config) {
-  const items = [];
-  const manifests = config.manifests;
-  for (const [_manifestName, manifestData] of Object.entries(manifests)) {
-    if (typeof manifestData === "object" && manifestData !== null) {
-      const data = manifestData;
-      if (Array.isArray(data.pages)) {
-        for (const page of data.pages) {
-          const normalized = toManifestItem2(page);
-          if (normalized) {
-            items.push(normalized);
-          }
-        }
+async function checkMachineHintsLiveEndpoints(config, verbose) {
+  const baseUrl = config.site.baseUrl.replace(/\/+$/, "");
+  const urls = [
+    config.machineHints?.robots ?? `${baseUrl}/robots.txt`,
+    config.machineHints?.sitemap ?? `${baseUrl}/sitemap.xml`,
+    config.machineHints?.llmsTxt ?? `${baseUrl}/llms.txt`,
+    config.machineHints?.llmsFullTxt ?? `${baseUrl}/llms-full.txt`
+  ];
+  const issues = [];
+  for (const url of urls) {
+    try {
+      if (verbose) {
+        printVerbose(`Live-checking ${url}`);
       }
-      if (Array.isArray(data)) {
-        for (const item of data) {
-          const normalized = toManifestItem2(item);
-          if (normalized) {
-            items.push(normalized);
-          }
-        }
+      const response = await fetch(url, {
+        method: "GET",
+        signal: AbortSignal.timeout(1e4)
+      });
+      if (!response.ok) {
+        issues.push({
+          path: url,
+          code: "invalid_url",
+          message: `Live check failed with HTTP ${response.status}`,
+          severity: "error"
+        });
       }
+    } catch (error) {
+      const message = error instanceof Error ? error.message : String(error);
+      issues.push({
+        path: url,
+        code: "invalid_url",
+        message: `Live check request failed: ${message}`,
+        severity: "error"
+      });
     }
   }
-  return items;
-}
-function toManifestItem2(value) {
-  if (typeof value !== "object" || value === null) {
-    return null;
-  }
-  const data = value;
-  const rawSlug = data.slug ?? data.path;
-  if (typeof rawSlug !== "string" || rawSlug.length === 0) {
-    return null;
-  }
-  const item = {
-    slug: rawSlug.startsWith("/") ? rawSlug : `/${rawSlug}`
-  };
-  if (typeof data.title === "string") {
-    item.title = data.title;
-  }
-  if (typeof data.description === "string") {
-    item.description = data.description;
-  }
-  if (Array.isArray(data.locales)) {
-    const locales = data.locales.filter((loc) => typeof loc === "string");
-    if (locales.length > 0) {
-      item.locales = locales;
-    }
-  }
-  if (typeof data.canonicalOverride === "string") {
-    item.canonicalOverride = data.canonicalOverride;
-  }
-  if (typeof data.publishedAt === "string") {
-    item.publishedAt = data.publishedAt;
-  }
-  if (typeof data.updatedAt === "string") {
-    item.updatedAt = data.updatedAt;
-  }
-  if (typeof data.priority === "number") {
-    item.priority = data.priority;
-  }
-  return item;
+  return issues;
 }
 var init_check = __esm({
   "src/cli/commands/check.ts"() {
@@ -2184,7 +2394,7 @@ program.command("generate").description("Generate llms.txt and llms-full.txt fro
   });
   process.exit(exitCode);
 });
-program.command("check").description("Validate generated llms.txt files against configuration").option("-c, --config <path>", "Path to config file").option("--fail-on <level>", "Fail on warnings (warn) or only errors (error)", "error").option("-v, --verbose", "Show detailed output", false).action(async (options) => {
+program.command("check").description("Validate generated llms.txt files against configuration").option("-c, --config <path>", "Path to config file").option("--fail-on <level>", "Fail on warnings (warn) or only errors (error)", "error").option("--check-machine-hints-live", "Check machine hint URLs (robots/sitemap/llms) over HTTP", false).option("-v, --verbose", "Show detailed output", false).action(async (options) => {
   const { checkCommand: checkCommand2 } = await Promise.resolve().then(() => (init_check(), check_exports));
   const failOn = options.failOn;
   if (failOn !== "warn" && failOn !== "error") {
@@ -2194,6 +2404,7 @@ program.command("check").description("Validate generated llms.txt files against
   const exitCode = await checkCommand2({
     config: options.config ?? program.opts().config,
     failOn,
+    checkMachineHintsLive: options.checkMachineHintsLive ?? false,
     verbose: options.verbose ?? program.opts().verbose ?? false
   });
   process.exit(exitCode);