seo 0.13.2 → 0.13.3

package/dist/index.d.ts

@@ -21539,6 +21539,12 @@ type AuditSummary = {
     capped_rules: string[];
     resolution: 'all' | 'crawled-pages-only';
   };
+  rendering?: {
+    mode: 'on' | 'off' | 'unavailable' | null;
+    rendered: number;
+    failed: number;
+    browser_ms: number;
+  };
   changes: {
     compared: boolean;
     new_urls: number;
@@ -21694,8 +21700,21 @@ type CommandDeps = {
   analytics?: AnalyticsAccess;
   audit?: AuditAccess;
   week?: () => Promise<WeekData | null>;
+  startAudit?: (site: string) => Promise<StartAudit>;
   now?: () => Date;
 };
+type StartAudit = {
+  status: 'added' | 'exists' | 'pending';
+} | {
+  status: 'checkout';
+  url: string;
+  price: number;
+} | {
+  status: 'full';
+  limit: number;
+} | {
+  status: 'payment_failed';
+};
 type PageMarkdown = {
   markdown: string;
   characters: number;

package/dist/index.js

@@ -1,5 +1,5 @@
 import { createRequire as __seoCreateRequire } from 'node:module'; const require = __seoCreateRequire(import.meta.url);
-import { n as version, t as load } from "./esm-Dmtr8qrP.js";
+import { n as version, t as load } from "./esm-hHlvm1W0.js";
 import { SEO_PROVIDER_API_VERSION } from "./provider-sdk.js";
 import crypto, { createHash, randomUUID, sign } from "node:crypto";
 import { z } from "zod";
@@ -1638,6 +1638,13 @@ async function failure(response, label, subject, options) {
 	const detail = (await response.text().catch(() => "")).slice(0, 300);
 	const status = response.status;
 	const what = subject ? `${label} for ${subject}` : label;
+	if (status === 400) {
+		let reason = detail;
+		try {
+			reason = JSON.parse(detail).error?.message ?? detail;
+		} catch {}
+		return new GoogleApiError("INVALID_INPUT", `${label} rejected the request: ${reason}`.trim(), status);
+	}
 	if (status === 401) return new GoogleApiError("AUTH_EXPIRED", `Google rejected the sign-in. ${options.signIn}`, status);
 	if (status === 403) return new GoogleApiError("ACCESS_DENIED", `Google denied access to ${what}. Check the Google account, its access to this property, and the granted scopes. ${detail}`.trim(), status);
 	if (status === 404) return new GoogleApiError("PROPERTY_NOT_FOUND", `Google could not find ${what}. Check the property and account access.`, status);
@@ -30359,7 +30366,7 @@ async function analyticsBeside(access, site, range, before, pages) {
 //#endregion
 //#region packages/core/src/commands/audit.ts
 const AUDIT_SECTION = "weekly-audit";
-const AUDIT_SECTION_TEXT = "What the weekly crawl of the site found (broken and redirected internal links, missing or duplicate titles, descriptions and H1s, noindex, canonicals to another page, sitemap URLs that do not return 200, thin pages), with what is new, back or resolved since the crawl before. Only for sites added to the weekly site audit beta.";
+const AUDIT_SECTION_TEXT = "What the weekly crawl of the site found (broken and redirected internal links, missing or duplicate titles, descriptions and H1s, noindex, canonicals to another page, sitemap URLs that do not return 200, thin pages, tags that JavaScript changes), with what is new, back or resolved since the crawl before. Only for sites added to the weekly site audit beta.";
 const AUDIT_RULES = {
 	link_to_error: "Internal links to a page that returns an error",
 	link_to_redirect: "Internal links to a redirect",
@@ -30372,7 +30379,8 @@ const AUDIT_RULES = {
 	canonical_elsewhere: "Pages whose canonical is another URL",
 	sitemap_url_not_ok: "Sitemap URLs that do not return 200",
 	sitemap_page_unlinked: "Sitemap pages no crawled page links to (heuristic: the crawl may not reach every page)",
-	thin_page: "Pages under 100 words (heuristic)"
+	thin_page: "Pages under 100 words (heuristic)",
+	js_changed_tags: "Pages whose raw HTML has another title, canonical or robots tag than the page after JavaScript runs"
 };
 const PAGE_SIZE$1 = 50;
 const SUMMARY_ROWS = 20;
@@ -30387,6 +30395,17 @@ function evidence(finding, origin) {
 	if (typeof details.location === "string") parts.push(`to ${relative$2(details.location, origin)}`);
 	if (typeof details.canonical === "string") parts.push(`canonical ${relative$2(details.canonical, origin)}`);
 	if (typeof details.words === "number") parts.push(`${details.words} words`);
+	if (details.raw && typeof details.raw === "object") {
+		const raw = details.raw;
+		for (const key of [
+			"title",
+			"canonical",
+			"robots"
+		]) if (key in raw) {
+			const value = raw[key];
+			parts.push(`raw ${key} ${typeof value === "string" ? key === "canonical" ? relative$2(value, origin) : value : "none"}`);
+		}
+	}
 	const linking = typeof details.linking_pages === "number";
 	const count = linking ? details.linking_pages : details.pages;
 	if (typeof count === "number") {
@@ -30417,6 +30436,14 @@ function ordered(findings, since) {
 	};
 	return [...findings].sort((a, b) => rank(a) - rank(b) || a.rule.localeCompare(b.rule) || (a.subject < b.subject ? -1 : a.subject > b.subject ? 1 : 0));
 }
+function renderingText(rendering) {
+	if (!rendering) return null;
+	const pages = (n) => n === 1 ? "1 page" : `${n} pages`;
+	if (rendering.mode === "on") return `JavaScript changed what some pages show, so the crawl rendered the parts of the site that need it (${pages(rendering.rendered)} rendered${rendering.failed ? `; ${pages(rendering.failed)} could not be rendered, so ${rendering.failed === 1 ? "it is" : "they are"} left out of the tag and text checks` : ""}).`;
+	if (rendering.mode === "off") return "JavaScript did not change the pages checked, so the crawl read raw HTML.";
+	if (rendering.mode === "unavailable") return "JavaScript rendering failed on the pages checked, so the crawl read raw HTML. If the site builds its pages with JavaScript, titles, canonicals and links may be missing here.";
+	return rendering.failed && !rendering.rendered ? "JavaScript rendering failed, so the crawl read raw HTML." : null;
+}
 function runAuditReport(input, snapshot, call) {
 	const site = input.site;
 	const detail = input.detail ?? "summary";
@@ -30465,7 +30492,8 @@ function runAuditReport(input, snapshot, call) {
 	const head = [
 		`Weekly site audit of ${site}, crawled ${day$1(snapshot.finished_at)}: ${coverage}.${pages.blocked ? ` ${pages.blocked === 1 ? "1 URL" : `${pages.blocked} URLs`} blocked by robots.txt ${pages.blocked === 1 ? "was" : "were"} not fetched.` : ""}`,
 		previous ? `Since the crawl of ${day$1(previous.finished_at)}: ${summary.findings.new} new, ${summary.findings.reopened} back, ${summary.findings.resolved} resolved. ${summary.findings.open} open now.` : `This is the first crawl, so every finding is new: ${summary.findings.open} open.`,
-		summary.findings.resolution === "crawled-pages-only" ? "Because the crawl is partial, a finding counts as resolved only when this crawl fetched its page again; a finding not seen here is not an all-clear." : null,
+		renderingText(summary.rendering),
+		summary.findings.resolution === "crawled-pages-only" ? summary.complete ? "Because some pages could not be rendered, a finding counts as resolved only when this crawl rendered its page; a finding not seen here is not an all-clear." : "Because the crawl is partial, a finding counts as resolved only when this crawl fetched its page again; a finding not seen here is not an all-clear." : null,
 		summary.findings.capped_rules.length ? `Capped at 1,000 findings each, so these counts are lower bounds: ${summary.findings.capped_rules.join(", ")}.` : null,
 		running
 	].filter(Boolean).join("\n");
@@ -32795,6 +32823,7 @@ const siteName = (property) => property.replace(/^sc-domain:|^https?:\/\//, "").
 const count = (n) => n.toLocaleString("en-US");
 const plural$1 = (n, one, many = `${one}s`) => `${count(n)} ${n === 1 ? one : many}`;
 const list = (items) => items.length <= 1 ? items[0] ?? "" : `${items.slice(0, -1).join(", ")} and ${items.at(-1)}`;
+const sentence = (text) => /[.!?]$/.test(text) ? text : `${text}.`;
 function path(url) {
 	try {
 		const parsed = new URL(url);
@@ -32817,60 +32846,58 @@ const percent = (site) => {
 	return Math.round(Math.abs(clicks.clicks - clicks.previous) / clicks.previous * 100);
 };
 function issues(site, startMs, endMs) {
-	const name = siteName(site.property);
 	const found = [];
+	if (site.index?.dropped) found.push({
+		rank: 0,
+		heading: `${plural$1(site.index.dropped, "page")} dropped`
+	});
+	const crawl = site.crawl;
+	const crawled = crawl !== null && crawl.finished_at >= startMs && crawl.finished_at < endMs;
+	if (crawled && crawl.status === "failed") found.push({
+		rank: 1,
+		heading: "crawl stopped early"
+	});
+	else if (!crawled && site.added_at < startMs) found.push({
+		rank: 1,
+		heading: "no crawl this week"
+	});
+	if (clicksChange(site) === "fell") found.push({
+		rank: 2,
+		heading: `clicks down ${percent(site)}%`
+	});
+	if (crawled && crawl.new_issues > 0) found.push({
+		rank: 3,
+		heading: plural$1(crawl.new_issues, "new issue")
+	});
+	return found;
+}
+function siteLines(site, startMs, endMs) {
+	const lines = [];
 	const index = site.index;
-	if (index?.dropped) {
+	if (!index) lines.push("Google index: no checks ran this week yet.");
+	else if (index.dropped) {
 		const paths = index.dropped_pages.map((page) => path(page.url));
 		const more = index.dropped - paths.length;
 		const shown = more > 0 ? `${paths.join(", ")} and ${plural$1(more, "more")}` : list(paths);
 		const reasons = new Set(index.dropped_pages.map((page) => page.coverage));
 		const reason = reasons.size === 1 ? [...reasons][0] : null;
-		found.push({
-			rank: 0,
-			heading: `${plural$1(index.dropped, "page")} dropped`,
-			line: `${shown} dropped out of Google’s index.${reason ? ` Google says: ${reason}.` : ""}`,
-			next: `check ${index.dropped === 1 ? "it" : "they"} still return${index.dropped === 1 ? "s" : ""} 200 and ${index.dropped === 1 ? "is" : "are"} not noindexed.`,
-			prompt: `Why did ${paths[0] ?? "a page"} on ${name} drop out of Google’s index this week, and how do I fix it?`
-		});
-	}
+		lines.push(`Google index: ${plural$1(index.dropped, "page")} dropped out (${shown}).${reason ? ` Google says: ${reason}.` : ""}${index.returned ? ` ${plural$1(index.returned, "page")} came back.` : ""} ${plural$1(index.checked, "page")} checked.`);
+	} else lines.push(`Google index: ${plural$1(index.checked, "page")} checked, none dropped${index.returned ? `, ${plural$1(index.returned, "page")} came back` : ""}.`);
 	const crawl = site.crawl;
 	const crawled = crawl !== null && crawl.finished_at >= startMs && crawl.finished_at < endMs;
-	if (crawled && crawl.status === "failed") found.push({
-		rank: 1,
-		heading: "crawl stopped early",
-		line: `The weekly crawl stopped after ${plural$1(crawl.pages, "page")}${crawl.error ? `: ${crawl.error}` : "."}`,
-		next: "check the site is up. The next weekly crawl tries again.",
-		prompt: `Why did the weekly crawl of ${name} stop early, and what should I check?`
-	});
-	else if (!crawled && site.added_at < startMs) found.push({
-		rank: 1,
-		heading: "no crawl this week",
-		line: `No weekly crawl finished this week${site.start_error ? `: ${site.start_error}` : "."}`,
-		next: "check the site is up. The crawl tries again each hour.",
-		prompt: `Why has the weekly crawl of ${name} not finished, and what should I check?`
-	});
-	if (clicksChange(site) === "fell") {
-		const clicks = site.clicks;
-		found.push({
-			rank: 2,
-			heading: `clicks down ${percent(site)}%`,
-			line: `Search clicks fell ${percent(site)}%, from ${count(clicks.previous)} to ${count(clicks.clicks)} (${weekLabel(clicks.start, clicks.end)}, against the 7 days before).`,
-			next: "ask your agent which pages and queries lost them.",
-			prompt: `Which pages and queries lost search clicks on ${name} in the last week, and why?`
-		});
-	}
-	if (crawled && crawl.new_issues > 0) {
+	if (crawled && crawl.status === "failed") lines.push(`Weekly crawl: stopped after ${plural$1(crawl.pages, "page")}${crawl.error ? `: ${sentence(crawl.error)}` : "."} It tries again next week.`);
+	else if (crawled) {
 		const rule = crawl.top_rule ? AUDIT_RULES[crawl.top_rule] : void 0;
-		found.push({
-			rank: 3,
-			heading: plural$1(crawl.new_issues, "new issue"),
-			line: `The weekly crawl found ${plural$1(crawl.new_issues, "new issue")}${rule ? `, most of them ${rule.charAt(0).toLowerCase()}${rule.slice(1)}` : ""}.`,
-			next: "ask your agent for the weekly audit of this site.",
-			prompt: `What did the weekly crawl of ${name} find this week, and what should I fix first?`
-		});
-	}
-	return found;
+		lines.push(crawl.new_issues ? `Weekly crawl: ${plural$1(crawl.pages, "page")}, ${plural$1(crawl.new_issues, "new issue")}${rule ? `, most of them ${rule.charAt(0).toLowerCase()}${rule.slice(1)}` : ""}.` : `Weekly crawl: ${plural$1(crawl.pages, "page")}, no new issues.`);
+	} else if (site.added_at < startMs) lines.push(`Weekly crawl: none finished this week${site.start_error ? `: ${sentence(site.start_error)}` : "."} It tries again each hour.`);
+	else lines.push("Weekly crawl: the first crawl is still running.");
+	const clicks = site.clicks;
+	if (clicks) {
+		const change = clicks.previous > 0 ? Math.round((clicks.clicks - clicks.previous) / clicks.previous * 100) : null;
+		const trend = change === null ? `from ${count(clicks.previous)}` : change === 0 ? "level with" : `${change > 0 ? "up" : "down"} ${Math.abs(change)}% on`;
+		lines.push(`Search clicks: ${count(clicks.clicks)} (${weekLabel(clicks.start, clicks.end)}), ${trend}${change === null ? "" : ` ${count(clicks.previous)}`} the 7 days before.`);
+	} else lines.push("Search clicks: not available this week.");
+	return lines;
 }
 function weekMarkdown(data) {
 	const startMs = Date.parse(`${data.start}T00:00:00Z`);
@@ -32878,43 +32905,21 @@ function weekMarkdown(data) {
 	const title = `# ${weekTitle(data)}`;
 	if (!data.sites.length) return `${title}\n\nNo site is in your weekly audit yet. Add one on the Sites tab: ${data.account}#sites`;
 	const byName = (a, b) => siteName(a.property).localeCompare(siteName(b.property)) || a.property.localeCompare(b.property);
-	const sites = [...data.sites].sort(byName);
-	const attention = [];
-	const quiet = [];
-	const waiting = [];
-	for (const site of sites) {
-		const found = issues(site, startMs, endMs);
-		if (found.length) attention.push({
-			site,
-			found
-		});
-		else if (!site.crawl || !site.index) waiting.push(site);
-		else quiet.push(site);
-	}
-	attention.sort((a, b) => a.found[0].rank - b.found[0].rank || byName(a.site, b.site));
-	const total = sites.length;
+	const rows = data.sites.map((site) => ({
+		site,
+		found: issues(site, startMs, endMs),
+		waiting: !site.crawl || !site.index
+	}));
+	const attention = rows.filter((row) => row.found.length);
+	const quiet = rows.filter((row) => !row.found.length && !row.waiting);
+	rows.sort((a, b) => (a.found[0]?.rank ?? Number.POSITIVE_INFINITY) - (b.found[0]?.rank ?? Number.POSITIVE_INFINITY) || byName(a.site, b.site));
+	const total = rows.length;
 	const out = [
 		title,
 		"",
-		attention.length ? total === 1 ? "Your site needs a look this week." : `${attention.length} of your ${total} sites ${attention.length === 1 ? "needs" : "need"} a look this week.` : quiet.length ? total === 1 ? "Your site had a quiet week." : quiet.length === total ? `All ${total} of your sites had a quiet week.` : `${quiet.length} of your ${total} sites had a quiet week.` : "Your sites are still on their first checks."
+		attention.length ? total === 1 ? "Your site had changes worth a look this week." : `${attention.length} of your ${total} sites had changes worth a look this week.` : quiet.length ? total === 1 ? "Your site had a quiet week." : quiet.length === total ? `All ${total} of your sites had a quiet week.` : `${quiet.length} of your ${total} sites had a quiet week.` : "Your sites are still on their first checks."
 	];
-	if (attention.length) {
-		out.push("", "## Needs attention");
-		for (const { site, found } of attention) {
-			const [first, ...rest] = found;
-			const also = rest.length ? ` Also: ${list(rest.map((item) => item.heading))}.` : "";
-			const rose = clicksChange(site) === "rose" ? ` Search clicks rose ${percent(site)}%.` : "";
-			out.push("", `### ${siteName(site.property)}: ${first.heading}`, "", `- ${first.line}${also}${rose}`, `- Next step: ${first.next}`, ...site.link ? [`- [View site](${site.link})`] : []);
-		}
-	}
-	if (quiet.length) {
-		const risen = quiet.filter((site) => clicksChange(site) === "rose");
-		out.push("", "## All quiet", "", `${list(quiet.map((site) => siteName(site.property)))}: no drops, no new issues.${risen.map((site) => ` Search clicks rose ${percent(site)}% on ${siteName(site.property)}.`).join("")}`);
-	}
-	if (waiting.length) out.push("", "## Not checked yet", "", `${list(waiting.map((site) => siteName(site.property)))}: the first crawl or Google checks are still running.`);
-	const first = attention[0];
-	const prompt = first ? first.found[0].prompt : `What are the best quick wins on ${siteName((quiet[0] ?? sites[0]).property)} this month?`;
-	out.push("", "## Ask your agent", "", "Paste this into your agent:", "", "```", prompt, "```");
+	for (const { site, found } of rows) out.push("", `## ${siteName(site.property)}${found.length ? `: ${list(found.map((item) => item.heading))}` : ""}`, "", ...siteLines(site, startMs, endMs).map((line) => `- ${line}`), ...site.link ? [`- [View site](${site.link})`] : []);
 	return out.join("\n");
 }
 async function runAllSites(deps, sites) {
@@ -32974,15 +32979,38 @@ const SHARED_TOOLS = [
 		name: "seo_sites",
 		command: "data",
 		title: "List sites",
-		description: "Lists the Search Console and Google Analytics properties the connected Google accounts can read, the Bing Webmaster site for each, and which research keys are set. Call it first.",
-		input: z.object({}),
-		run: listSites$1
+		description: "Lists the Search Console and Google Analytics properties the connected Google accounts can read, the Bing Webmaster site for each, and which research keys are set. Call it first. With add: starts the weekly site audit of that site on the hosted server (a weekly crawl, daily Google index checks, a Monday email and webhook with what changed, and alerts when a page drops out of Google's index). A site that needs paying for returns a Stripe Checkout link: give it to the person to open, and the audit starts once they pay.",
+		input: z.object({ add: site.optional().describe("Optional. A site to add to the weekly site audit. Only when the person asks for it: a site can cost money.") }),
+		run: async (surface, input) => {
+			if (!input.add) return listSites$1(surface);
+			const start = surface.deps.startAudit;
+			if (!start) throw new CommandError("INVALID_INPUT", "The weekly site audit runs on the hosted server only. Connect https://seoskill.dev/mcp and call seo_sites with add there.");
+			const property = await surface.resolveSite(input.add);
+			const result = await start(property);
+			return {
+				text: {
+					added: `Added ${property} to the weekly site audit. Its first crawl and index checks have started; ask ${surface.deps.call("report", {
+						site: property,
+						section: AUDIT_SECTION
+					})} for the findings once the crawl is done.`,
+					exists: `${property} is already in the weekly site audit.`,
+					pending: "The payment went through and the plan is starting. Call seo_sites with add again in a minute.",
+					checkout: result.status === "checkout" ? `${property} needs a plan: $${result.price} a month per site, cancel at any time. Give the person this Stripe Checkout link to open and pay: ${result.url}\nThe audit of ${property} starts once they pay. Call seo_sites with add again afterwards to confirm.` : "",
+					full: result.status === "full" ? `This account already audits ${result.limit} sites, the most it can. Remove one on the account page first.` : "",
+					payment_failed: `Stripe could not take the payment for another site, so ${property} was not added. The person can check their card under Manage billing on the account page.`
+				}[result.status],
+				data: {
+					property,
+					...result
+				}
+			};
+		}
 	}),
 	tool({
 		name: "seo_report",
 		command: "report",
 		title: "Search Console report",
-		description: `Start here. Without site: a short summary of every site in the person's weekly site audit (beta) for last week, with what needs a look and the next step for each, ending with the site to ask about next; an account outside that beta gets its site list to pick from. With site: the main report for that Search Console property, the overview plus each section's row count and top rows. With site and section: that section's rows; page sections also check their top 20 URLs live. Sections:\n${Object.entries(SECTIONS).map(([id, text]) => `- ${id}: ${text}`).join("\n")}`,
+		description: `Start here. Without site: a short summary of every site in the person's weekly site audit (beta) for last week, with each site's Google index, weekly crawl and search clicks, sites with changes first, ending with the site to ask about next; an account outside that beta gets its site list to pick from. With site: the main report for that Search Console property, the overview plus each section's row count and top rows. With site and section: that section's rows; page sections also check their top 20 URLs live. Sections:\n${Object.entries(SECTIONS).map(([id, text]) => `- ${id}: ${text}`).join("\n")}`,
 		input: z.object({
 			site: site.optional().describe("Leave it out for the summary of every audited site (weekly site audit beta). Otherwise a Search Console property or site, such as sc-domain:example.com, https://example.com/ or example.com."),
 			section: z.enum(SECTION_IDS).optional().describe("Optional, needs site. Leave it out first; then ask for one section."),