pagesight 0.17.0 → 0.19.0

package/src/tools/search/analytics.ts

@@ -0,0 +1,266 @@
+import { querySearchAnalytics, type SearchAnalyticsFilter, type SearchAnalyticsResponse } from "../../providers/gsc.js";
+import { pacificDaysAgo } from "../../shared/dates.js";
+import { type SearchOptions } from "./schema.js";
+import { textResult } from "./result.js";
+interface Totals {
+  clicks: number;
+  impressions: number;
+  ctr: number;
+  position: number;
+}
+
+function computeTotals(rows: SearchAnalyticsResponse["rows"]): Totals {
+  const r = rows ?? [];
+  const clicks = r.reduce((sum, row) => sum + row.clicks, 0);
+  const impressions = r.reduce((sum, row) => sum + row.impressions, 0);
+  const ctr = impressions > 0 ? clicks / impressions : 0;
+  const position = impressions > 0 ? r.reduce((sum, row) => sum + row.position * row.impressions, 0) / impressions : 0;
+  return { clicks, impressions, ctr, position };
+}
+
+function pctChange(current: number, previous: number): string {
+  if (previous === 0) return current > 0 ? `+${current} (new)` : "0%";
+  const change = ((current - previous) / previous) * 100;
+  return `${change > 0 ? "+" : ""}${change.toFixed(1)}%`;
+}
+
+function formatPerformance(
+  siteUrl: string,
+  result: SearchAnalyticsResponse,
+  dimensions: string[],
+  startDate: string,
+  endDate: string,
+): string {
+  const rows = result.rows ?? [];
+  const lines: string[] = [
+    `=== Search Performance: ${siteUrl} ===`,
+    `Period: ${startDate} to ${endDate}`,
+    `Dimensions: ${dimensions.join(", ")}`,
+    `Aggregation: ${result.responseAggregationType ?? "auto"}`,
+    `Results: ${rows.length}`,
+    "",
+  ];
+
+  if (result.metadata) {
+    if (result.metadata.first_incomplete_date)
+      lines.push(`Data incomplete from: ${result.metadata.first_incomplete_date}`);
+    if (result.metadata.first_incomplete_hour)
+      lines.push(`Hourly data incomplete from: ${result.metadata.first_incomplete_hour}`);
+    lines.push("");
+  }
+
+  if (rows.length === 0) {
+    lines.push("No data found for this period and filters.");
+    return lines.join("\n");
+  }
+
+  const totals = computeTotals(rows);
+
+  if (totals.impressions < 10) {
+    lines.push("⚠ Low data volume — trends may not be meaningful.", "");
+  }
+
+  lines.push(
+    `--- Returned-row summary (${rows.length} rows; not property totals) ---`,
+    "API top-row limits and query privacy exclusions may omit data.",
+    "",
+    `Clicks: ${totals.clicks.toLocaleString()}`,
+    `Impressions: ${totals.impressions.toLocaleString()}`,
+    `Avg CTR: ${(totals.ctr * 100).toFixed(1)}%`,
+    `Avg Position: ${totals.position.toFixed(1)}`,
+    "",
+    "--- Top Results ---",
+    "",
+  );
+
+  const top = rows.slice(0, 25);
+  for (const row of top) {
+    const keys = row.keys.map((k, i) => `${dimensions[i] ?? "key"}=${k}`).join(" | ");
+    lines.push(`${keys}`);
+    lines.push(
+      `  Clicks: ${row.clicks} | Impressions: ${row.impressions} | CTR: ${(row.ctr * 100).toFixed(1)}% | Position: ${row.position.toFixed(1)}`,
+    );
+  }
+
+  if (rows.length > 25) {
+    lines.push("", `... and ${rows.length - 25} more rows`);
+  }
+
+  return lines.join("\n");
+}
+
+export function formatComparison(
+  siteUrl: string,
+  current: SearchAnalyticsResponse,
+  previous: SearchAnalyticsResponse,
+  dimensions: string[],
+  currentStart: string,
+  currentEnd: string,
+  previousStart: string,
+  previousEnd: string,
+): string {
+  const curRows = current.rows ?? [];
+  const prevRows = previous.rows ?? [];
+  const cur = computeTotals(curRows);
+  const prev = computeTotals(prevRows);
+
+  const lines: string[] = [
+    `=== Search Performance: ${siteUrl} ===`,
+    `Current:  ${currentStart} to ${currentEnd}`,
+    `Previous: ${previousStart} to ${previousEnd}`,
+    `Dimensions: ${dimensions.join(", ")}`,
+    "",
+  ];
+
+  if (cur.impressions < 10 && prev.impressions < 10) {
+    lines.push("⚠ Low data volume — trends may not be meaningful.", "");
+  }
+
+  lines.push("--- Returned-row summary (not property totals) ---", "");
+  lines.push("Query privacy exclusions, top-row limits and aggregation may omit data.");
+  lines.push("Position changes can reflect a different mix of impressions, not a ranking change.", "");
+  lines.push("             Current     Previous    Change");
+  lines.push(
+    `Clicks:      ${String(cur.clicks.toLocaleString()).padEnd(12)} ${String(prev.clicks.toLocaleString()).padEnd(12)} ${pctChange(cur.clicks, prev.clicks)}`,
+  );
+  lines.push(
+    `Impressions: ${String(cur.impressions.toLocaleString()).padEnd(12)} ${String(prev.impressions.toLocaleString()).padEnd(12)} ${pctChange(cur.impressions, prev.impressions)}`,
+  );
+  const curCtr = `${(cur.ctr * 100).toFixed(1)}%`;
+  const prevCtr = `${(prev.ctr * 100).toFixed(1)}%`;
+  lines.push(`Avg CTR:     ${curCtr.padEnd(12)} ${prevCtr.padEnd(12)} ${((cur.ctr - prev.ctr) * 100).toFixed(1)}pp`);
+  const posStatus =
+    prev.position === 0 && cur.position > 0
+      ? "new"
+      : cur.position < prev.position
+        ? "improved"
+        : cur.position > prev.position
+          ? "regressed"
+          : "stable";
+  lines.push(
+    `Avg Position:${String(cur.position.toFixed(1)).padEnd(13)} ${String(prev.position.toFixed(1)).padEnd(12)} ${posStatus} (${(cur.position - prev.position).toFixed(1)})`,
+  );
+  lines.push("");
+
+  // Build lookup for previous period rows
+  const prevMap = new Map<string, (typeof prevRows)[0]>();
+  for (const row of prevRows) {
+    prevMap.set(JSON.stringify(row.keys), row);
+  }
+
+  // Find biggest movers (by click change)
+  const movers: Array<{ keys: string[]; curClicks: number; prevClicks: number; curPos: number; prevPos: number }> = [];
+  for (const row of curRows) {
+    const key = JSON.stringify(row.keys);
+    const prevRow = prevMap.get(key);
+    if (!prevRow) continue;
+    movers.push({
+      keys: row.keys,
+      curClicks: row.clicks,
+      prevClicks: prevRow.clicks,
+      curPos: row.position,
+      prevPos: prevRow.position,
+    });
+  }
+
+  lines.push("Movers include only rows observed in both periods. Missing rows are unknown, not zero.", "");
+
+  movers.sort((a, b) => Math.abs(b.curClicks - b.prevClicks) - Math.abs(a.curClicks - a.prevClicks));
+
+  const improved = movers.filter((m) => m.curClicks > m.prevClicks).slice(0, 10);
+  const regressed = movers.filter((m) => m.curClicks < m.prevClicks).slice(0, 10);
+
+  if (improved.length > 0) {
+    lines.push("--- Improved ---", "");
+    for (const m of improved) {
+      const keys = m.keys.map((k, i) => `${dimensions[i] ?? "key"}=${k}`).join(" | ");
+      const posChange = m.prevPos > 0 ? ` | Position: ${m.prevPos.toFixed(1)} → ${m.curPos.toFixed(1)}` : "";
+      lines.push(`${keys}`);
+      lines.push(`  Clicks: ${m.prevClicks} → ${m.curClicks} (${pctChange(m.curClicks, m.prevClicks)})${posChange}`);
+    }
+    lines.push("");
+  }
+
+  lines.push("--- Regressed ---", "");
+  if (regressed.length > 0) {
+    for (const m of regressed) {
+      const keys = m.keys.map((k, i) => `${dimensions[i] ?? "key"}=${k}`).join(" | ");
+      const posChange = m.prevPos > 0 ? ` | Position: ${m.prevPos.toFixed(1)} → ${m.curPos.toFixed(1)}` : "";
+      lines.push(`${keys}`);
+      lines.push(`  Clicks: ${m.prevClicks} → ${m.curClicks} (${pctChange(m.curClicks, m.prevClicks)})${posChange}`);
+    }
+  } else {
+    lines.push("(none)");
+  }
+  lines.push("");
+
+  return lines.join("\n");
+}
+
+export async function runAnalytics(options: SearchOptions) {
+  const {
+    site_url,
+    start_date,
+    end_date,
+    dimensions,
+    search_type,
+    data_state,
+    aggregation_type,
+    filters,
+    row_limit,
+    start_row,
+    compare,
+  } = options;
+  if (!site_url) return textResult("Error: site_url is required for analytics.");
+
+  const startDate = start_date ?? pacificDaysAgo(28);
+  const endDate = end_date ?? pacificDaysAgo(3);
+
+  if (start_date && Number.isNaN(new Date(start_date).getTime())) {
+    return textResult(`Error: invalid start_date "${start_date}". Use YYYY-MM-DD format.`);
+  }
+  if (end_date && Number.isNaN(new Date(end_date).getTime())) {
+    return textResult(`Error: invalid end_date "${end_date}". Use YYYY-MM-DD format.`);
+  }
+  const dims = dimensions ?? ["query", "page"];
+
+  const filterGroups =
+    filters && filters.length > 0
+      ? [{ groupType: "and" as const, filters: filters as SearchAnalyticsFilter[] }]
+      : undefined;
+
+  const queryOpts = {
+    startDate,
+    endDate,
+    dimensions: dims,
+    type: search_type,
+    dataState: data_state,
+    aggregationType: aggregation_type,
+    rowLimit: row_limit ?? 1000,
+    startRow: start_row,
+    dimensionFilterGroups: filterGroups,
+  };
+
+  if (compare) {
+    // Calculate previous period of equal length
+    const start = new Date(startDate);
+    const end = new Date(endDate);
+    const durationMs = end.getTime() - start.getTime();
+    const prevEnd = new Date(start.getTime() - 1 * 24 * 60 * 60 * 1000); // day before current start
+    const prevStart = new Date(prevEnd.getTime() - durationMs);
+    const prevStartDate = prevStart.toISOString().split("T")[0];
+    const prevEndDate = prevEnd.toISOString().split("T")[0];
+
+    const [current, previous] = await Promise.all([
+      querySearchAnalytics(site_url, queryOpts),
+      querySearchAnalytics(site_url, { ...queryOpts, startDate: prevStartDate, endDate: prevEndDate }),
+    ]);
+
+    return textResult(
+      formatComparison(site_url, current, previous, dims, startDate, endDate, prevStartDate, prevEndDate),
+    );
+  }
+
+  const result = await querySearchAnalytics(site_url, queryOpts);
+  return textResult(formatPerformance(site_url, result, dims, startDate, endDate));
+}

package/src/tools/search/coverage.ts

@@ -0,0 +1,247 @@
+import { listSitemaps } from "../../providers/gsc.js";
+import { type SearchOptions } from "./schema.js";
+import { humanizeState, type InspectionSummary, inspectSingle } from "./inspection.js";
+import { fetchSitemap, sampleUrls } from "./sitemap-sampling.js";
+import { textResult } from "./result.js";
+const COVERAGE_FILTERS = [
+  "not_indexed",
+  "server_error",
+  "redirect",
+  "soft_404",
+  "blocked",
+  "duplicate",
+  "discovered",
+  "crawled_not_indexed",
+] as const;
+
+type CoverageFilter = (typeof COVERAGE_FILTERS)[number];
+
+export function matchesCoverageFilter(result: InspectionSummary, filter: CoverageFilter): boolean {
+  if (result.error) return false;
+  const cs = result.coverageState.toLowerCase();
+  const pf = result.pageFetchState.toLowerCase();
+  switch (filter) {
+    case "not_indexed":
+      return result.verdict !== "PASS";
+    case "server_error":
+      return cs.includes("server error") || pf === "server_error";
+    case "redirect":
+      return cs.includes("redirect") || pf.includes("redirect");
+    case "soft_404":
+      return cs.includes("soft 404") || pf === "soft_404";
+    case "blocked":
+      return cs.includes("blocked");
+    case "duplicate":
+      return cs.includes("duplicate");
+    case "discovered":
+      return cs.includes("discovered");
+    case "crawled_not_indexed":
+      return cs.includes("crawled - currently not indexed");
+    default:
+      return false;
+  }
+}
+
+function filterNameForState(state: string): string | null {
+  const s = state.toLowerCase();
+  if (s.includes("server error")) return "server_error";
+  if (s.includes("blocked")) return "blocked";
+  if (s.includes("redirect")) return "redirect";
+  if (s.includes("soft 404")) return "soft_404";
+  if (s.includes("duplicate")) return "duplicate";
+  if (s.includes("discovered - currently not indexed")) return "discovered";
+  if (s.includes("crawled - currently not indexed")) return "crawled_not_indexed";
+  return null;
+}
+
+export function formatSampleResults(
+  siteUrl: string,
+  sitemapUrl: string,
+  totalUrls: number,
+  results: InspectionSummary[],
+): string {
+  const lines: string[] = [
+    `=== Sample Inspection: ${siteUrl} ===`,
+    `Sitemap: ${sitemapUrl} (${totalUrls.toLocaleString()} URLs)`,
+    `Sampled: ${results.length}`,
+    "",
+  ];
+
+  // Summary
+  const verdictCounts: Record<string, number> = {};
+  const coverageCounts: Record<string, number> = {};
+  const fetchCounts: Record<string, number> = {};
+  let indexed = 0;
+  let errors = 0;
+
+  for (const r of results) {
+    if (r.error) {
+      errors++;
+      continue;
+    }
+    verdictCounts[r.verdict] = (verdictCounts[r.verdict] ?? 0) + 1;
+    coverageCounts[r.coverageState] = (coverageCounts[r.coverageState] ?? 0) + 1;
+    fetchCounts[r.pageFetchState] = (fetchCounts[r.pageFetchState] ?? 0) + 1;
+    if (r.verdict === "PASS") indexed++;
+  }
+
+  const inspected = results.length - errors;
+  lines.push("--- Inspected sample ---", "");
+  lines.push(`Indexed: ${indexed}/${inspected}`);
+
+  if (indexed < inspected) {
+    lines.push(`Not indexed: ${inspected - indexed}/${inspected}`);
+    for (const [state, count] of Object.entries(coverageCounts)) {
+      if (state !== "Submitted and indexed" && state !== "Indexing allowed") {
+        lines.push(`  ${state}: ${count}`);
+      }
+    }
+  }
+
+  // Page fetch issues
+  const fetchIssues = Object.entries(fetchCounts).filter(([s]) => s !== "SUCCESSFUL");
+  if (fetchIssues.length > 0) {
+    lines.push("");
+    lines.push("Page fetch issues:");
+    for (const [state, count] of fetchIssues) {
+      lines.push(`  ${humanizeState(state)}: ${count}`);
+    }
+  }
+
+  if (errors > 0) {
+    lines.push(`\nInspection errors: ${errors}`);
+  }
+
+  // Individual results
+  lines.push("", "--- Details ---", "");
+  for (let i = 0; i < results.length; i++) {
+    const r = results[i];
+    lines.push(`${i + 1}. ${r.url}`);
+    if (r.error) {
+      lines.push(`   Error: ${r.error}`);
+    } else {
+      lines.push(`   Verdict: ${r.verdict}`);
+      lines.push(`   Coverage: ${r.coverageState}`);
+      lines.push(`   Page fetch: ${humanizeState(r.pageFetchState)}`);
+      if (r.robotsTxtState !== "ALLOWED") lines.push(`   Robots.txt: ${humanizeState(r.robotsTxtState)}`);
+      if (r.indexingState !== "INDEXING_ALLOWED") lines.push(`   Indexing: ${humanizeState(r.indexingState)}`);
+      if (r.lastCrawlTime) lines.push(`   Last crawled: ${r.lastCrawlTime}`);
+      if (r.googleCanonical && r.googleCanonical !== r.url) {
+        lines.push(`   Google canonical: ${r.googleCanonical}`);
+      }
+    }
+    lines.push("");
+  }
+
+  return lines.join("\n");
+}
+
+function formatCoverage(siteUrl: string, sitemapUrl: string, totalUrls: number, results: InspectionSummary[]): string {
+  const inspected = results.filter((r) => !r.error);
+  const errors = results.filter((r) => r.error);
+  const indexed = inspected.filter((r) => r.verdict === "PASS");
+  const notIndexed = inspected.filter((r) => r.verdict !== "PASS");
+
+  const lines: string[] = [
+    `=== Index Coverage: ${siteUrl} ===`,
+    `Sitemap: ${sitemapUrl} (${totalUrls.toLocaleString()} URLs)`,
+    `Inspected: ${results.length}`,
+    "",
+  ];
+
+  const total = inspected.length;
+  if (total === 0) {
+    lines.push("No URLs could be inspected.");
+    if (errors.length > 0) {
+      lines.push("");
+      for (const e of errors) lines.push(`  ${e.url}: ${e.error}`);
+    }
+    return lines.join("\n");
+  }
+
+  // Group not-indexed by coverageState
+  const issueGroups: Record<string, InspectionSummary[]> = {};
+  for (const r of notIndexed) {
+    const key = r.coverageState;
+    if (!issueGroups[key]) issueGroups[key] = [];
+    issueGroups[key].push(r);
+  }
+  const sortedIssues = Object.entries(issueGroups).sort((a, b) => b[1].length - a[1].length);
+
+  // Breakdown
+  lines.push("--- Breakdown ---", "");
+  lines.push(`Indexed: ${indexed.length}/${total} (${Math.round((indexed.length / total) * 100)}%)`);
+  lines.push(`Not indexed: ${notIndexed.length}/${total} (${Math.round((notIndexed.length / total) * 100)}%)`);
+  for (const [state, urls] of sortedIssues) {
+    lines.push(`  ${state}: ${urls.length}`);
+  }
+  if (errors.length > 0) {
+    lines.push(`Inspection errors: ${errors.length}`);
+  }
+
+  lines.push("", "These verdicts describe only the inspected sample, not site-wide coverage.");
+
+  // Example URLs per issue
+  if (sortedIssues.length > 0) {
+    lines.push("", "--- URLs by Issue ---", "");
+    for (const [state, urls] of sortedIssues) {
+      lines.push(`${state}:`);
+      for (const u of urls.slice(0, 3)) {
+        lines.push(`  ${u.url}`);
+        if (u.lastCrawlTime) lines.push(`    Last crawled: ${u.lastCrawlTime}`);
+        if (u.pageFetchState !== "SUCCESSFUL") lines.push(`    Fetch: ${humanizeState(u.pageFetchState)}`);
+      }
+      if (urls.length > 3) lines.push(`  ... and ${urls.length - 3} more in sample`);
+      lines.push("");
+    }
+  }
+
+  // Suggest drill-down
+  const filterSuggestions = sortedIssues.map(([state]) => filterNameForState(state)).filter(Boolean);
+  if (filterSuggestions.length > 0) {
+    lines.push(`Drill deeper: search(action='sample', filter='...') to investigate specific issues.`);
+    lines.push(`  Available filters: ${filterSuggestions.join(", ")}`);
+  }
+
+  return lines.join("\n");
+}
+
+export async function runCoverage(options: SearchOptions) {
+  const { site_url, sitemap_url, sample_size, sample_strategy } = options;
+  if (!site_url) return textResult("Error: site_url is required for coverage.");
+
+  const count = sample_size ?? 20;
+  const strategy = sample_strategy ?? "spread";
+
+  // Discover sitemap URL if not provided
+  let resolvedSitemapUrl = sitemap_url;
+  if (!resolvedSitemapUrl) {
+    const sitemaps = await listSitemaps(site_url);
+    if (sitemaps.length === 0) {
+      return textResult(`No sitemaps found for ${site_url} in GSC. Provide a sitemap_url directly.`);
+    }
+    const nonIndex = sitemaps.find((s) => !s.isSitemapsIndex);
+    resolvedSitemapUrl = (nonIndex ?? sitemaps[0]).path;
+  }
+
+  let parsed = await fetchSitemap(resolvedSitemapUrl);
+
+  if (parsed.isSitemapIndex && parsed.childSitemaps.length > 0) {
+    const childUrl = parsed.childSitemaps[0];
+    parsed = await fetchSitemap(childUrl);
+    resolvedSitemapUrl = `${resolvedSitemapUrl} → ${childUrl}`;
+  }
+
+  if (parsed.urls.length === 0) {
+    return textResult(`Sitemap ${resolvedSitemapUrl} contains no URLs.`);
+  }
+
+  const sampled = sampleUrls(parsed.urls, count, strategy);
+
+  const results: InspectionSummary[] = [];
+  for (const u of sampled) {
+    results.push(await inspectSingle(u, site_url));
+  }
+
+  return textResult(formatCoverage(site_url, resolvedSitemapUrl, parsed.urls.length, results));
+}

package/src/tools/search/gaps.ts

@@ -0,0 +1,129 @@
+import { querySearchAnalytics, type SearchAnalyticsFilter } from "../../providers/gsc.js";
+import { pacificDaysAgo } from "../../shared/dates.js";
+import { type SearchOptions } from "./schema.js";
+import { textResult } from "./result.js";
+function extractPageText(html: string): string {
+  // Extract text from HTML; this does not check rendered visibility.
+  let text = html;
+  text = text.replace(/<script[^>]*>[\s\S]*?<\/script>/gi, " ");
+  text = text.replace(/<style[^>]*>[\s\S]*?<\/style>/gi, " ");
+  text = text.replace(/<[^>]+>/g, " ");
+  // Decode common HTML entities
+  text = text
+    .replace(/&amp;/g, "&")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&#39;/g, "'")
+    .replace(/&nbsp;/g, " ");
+  // Normalize whitespace
+  text = text.replace(/\s+/g, " ").trim().toLowerCase();
+  return text;
+}
+
+function formatGapAnalysis(
+  pageUrl: string,
+  siteUrl: string,
+  queries: Array<{ query: string; clicks: number; impressions: number; ctr: number; position: number }>,
+  pageText: string,
+): string {
+  const lines: string[] = [`=== Keyword Gap Analysis: ${pageUrl} ===`, `Property: ${siteUrl}`, ""];
+
+  if (queries.length === 0) {
+    lines.push("No search queries found for this page.");
+    return lines.join("\n");
+  }
+
+  const onPage: typeof queries = [];
+  const gaps: typeof queries = [];
+
+  for (const q of queries) {
+    const words = q.query.toLowerCase().split(/\s+/);
+    const allPresent = words.every((w) => pageText.includes(w));
+    if (allPresent) {
+      onPage.push(q);
+    } else {
+      gaps.push(q);
+    }
+  }
+
+  lines.push(`Queries analyzed: ${queries.length}`);
+  lines.push(`On page: ${onPage.length}`);
+  lines.push(`Gaps (not on page): ${gaps.length}`);
+  lines.push("");
+
+  if (gaps.length > 0) {
+    lines.push("--- Keyword Gaps (queries with impressions but missing from page) ---", "");
+    gaps.sort((a, b) => b.impressions - a.impressions);
+    for (const q of gaps.slice(0, 20)) {
+      const ctr = (q.ctr * 100).toFixed(1);
+      lines.push(
+        `  "${q.query}" — ${q.impressions} imp, ${q.clicks} clicks, CTR ${ctr}%, pos ${q.position.toFixed(1)}`,
+      );
+    }
+    if (gaps.length > 20) lines.push(`  ... and ${gaps.length - 20} more`);
+    lines.push("");
+  }
+
+  if (onPage.length > 0) {
+    lines.push("--- Already Targeted (query found on page) ---", "");
+    onPage.sort((a, b) => b.impressions - a.impressions);
+    for (const q of onPage.slice(0, 10)) {
+      const ctr = (q.ctr * 100).toFixed(1);
+      lines.push(
+        `  "${q.query}" — ${q.impressions} imp, ${q.clicks} clicks, CTR ${ctr}%, pos ${q.position.toFixed(1)}`,
+      );
+    }
+    if (onPage.length > 10) lines.push(`  ... and ${onPage.length - 10} more`);
+    lines.push("");
+  }
+
+  const totalGapImpressions = gaps.reduce((sum, q) => sum + q.impressions, 0);
+  const totalImpressions = queries.reduce((sum, q) => sum + q.impressions, 0);
+  if (totalImpressions > 0) {
+    const gapPct = Math.round((totalGapImpressions / totalImpressions) * 100);
+    lines.push(`${gapPct}% of impressions come from queries not on the page — potential content opportunity.`);
+  }
+
+  return lines.join("\n");
+}
+
+export async function runGaps(options: SearchOptions) {
+  const { site_url, url, start_date, end_date, row_limit } = options;
+  if (!url) return textResult("Error: url is required for gap analysis.");
+  if (!site_url) return textResult("Error: site_url is required for gap analysis.");
+
+  // Fetch page content and GSC queries in parallel
+  const [pageRes, analyticsRes] = await Promise.all([
+    fetch(url, {
+      headers: { "User-Agent": "Mozilla/5.0 (compatible; Googlebot/2.1)", Accept: "text/html" },
+      redirect: "follow",
+    }),
+    querySearchAnalytics(site_url, {
+      startDate: start_date ?? pacificDaysAgo(28),
+      endDate: end_date ?? pacificDaysAgo(3),
+      dimensions: ["query"],
+      rowLimit: row_limit ?? 500,
+      dimensionFilterGroups: [
+        {
+          groupType: "and" as const,
+          filters: [{ dimension: "page", operator: "equals", expression: url } as SearchAnalyticsFilter],
+        },
+      ],
+    }),
+  ]);
+
+  if (!pageRes.ok) return textResult(`Error fetching ${url}: HTTP ${pageRes.status}`);
+  const html = await pageRes.text();
+  const pageText = extractPageText(html);
+
+  const queries = (analyticsRes.rows ?? []).map((row) => ({
+    query: row.keys[0],
+    clicks: row.clicks,
+    impressions: row.impressions,
+    ctr: row.ctr,
+    position: row.position,
+  }));
+
+  return textResult(formatGapAnalysis(url, site_url, queries, pageText));
+}