Merge pull request #12 from sampleXbro/develop

Fix website SEO deployment and link handling

Author: sampleXbro

.github/workflows/deploy-website.yml

@@ -7,8 +7,9 @@
 #
 # Live site: https://samplexbro.github.io/agentsmesh/ (see website/astro.config.mjs `site` + `base`).
 #
-# SEO: set repository variable DEPLOY_SITE_URL to the exact origin you want indexed
-# (HTTPS, no trailing slash), e.g. https://samplexbro.github.io or your custom domain.
+# SEO: set repository variable DEPLOY_SITE_URL to the exact public site URL
+# you want indexed, e.g. https://samplexbro.github.io/agentsmesh/ or
+# https://docs.agentsmesh.dev/. Custom-domain builds emit CNAME automatically.
 # Configure DNS or CDN to 301 the non-canonical hostname (www ↔ apex) to that URL.
 name: Deploy Website
 
@@ -55,6 +56,10 @@ jobs:
         working-directory: website
         run: pnpm install --frozen-lockfile
 
+      - name: Run website tests
+        working-directory: website
+        run: pnpm test
+
       - name: Build website
         working-directory: website
         env:

tasks/todo.md

@@ -1,3 +1,59 @@
+# Website SEO hardening pass
+
+- [x] Reproduce the current website SEO artifact/output state and confirm root causes for the reported issues
+- [x] Add failing tests first for deploy URL parsing and generated SEO artifacts (`robots.txt`, canonical URL/base handling, custom-domain `CNAME`)
+- [x] Implement the minimal website config/build changes to fix the reproducible issues
+- [x] Improve crawlable homepage copy to address the low text-to-code warning without changing product meaning
+- [x] Run website verification plus post-feature QA and append review notes
+
+## Review (Website SEO hardening pass)
+
+- Changes implemented:
+  - replaced the website’s hardcoded GitHub Pages path assumptions with a single resolved `DEPLOY_SITE_URL` contract that derives origin, base path, public URLs, and optional custom-domain `CNAME` output from one source of truth
+  - expanded the SEO build integration so it always writes `robots.txt` and writes `CNAME` automatically for custom-domain deployments
+  - converted hardcoded `/agentsmesh/...` doc links to base-agnostic relative links so the docs build works both on the current GitHub Pages project URL and on a root custom domain
+  - added more crawlable homepage copy to improve the low text-to-code ratio without changing the product positioning
+  - added website unit tests and wired them into `website/package.json` plus the deploy workflow so the SEO behavior is enforced in automation
+- Tests added:
+  - `website/site-url.test.mjs`
+  - `website/integrations/seo-robots.test.mjs`
+- Verification:
+  - `node --test website/site-url.test.mjs website/integrations/seo-robots.test.mjs`
+  - `node ./node_modules/astro/astro.js build` (in `website/`)
+  - `DEPLOY_SITE_URL=https://docs.agentsmesh.dev/ node ./node_modules/astro/astro.js build` (in `website/`)
+  - `rg -n 'href="/agentsmesh/|src="/agentsmesh/|https://docs.agentsmesh.dev/agentsmesh' website/dist -g '*.html'`
+  - inspected generated `website/dist/robots.txt` and custom-domain `website/dist/CNAME`
+- QA Report — Website SEO hardening pass
+
+### Acceptance Criteria
+
+| Criterion | Covered by test? | Status |
+| --- | --- | --- |
+| Deploy URL handling supports both GitHub Pages project paths and root custom domains | `website/site-url.test.mjs`, default/custom Astro builds | OK |
+| SEO artifact generation emits the correct `robots.txt` and optional `CNAME` from the same deploy URL | `website/integrations/seo-robots.test.mjs`, custom-domain Astro build | OK |
+| Internal docs links remain valid when the website base path changes | default/custom Astro builds plus no `/agentsmesh/` matches in custom-domain HTML output | OK |
+| Homepage ships more crawlable explanatory copy to help the text-to-code warning | updated `website/src/content/docs/index.mdx`, verified in Astro build output | OK |
+| Website deploy automation exercises the new SEO tests before building | `website/package.json`, `.github/workflows/deploy-website.yml` | OK |
+
+### Edge Cases
+
+| Scenario | Covered? | Test location |
+| --- | --- | --- |
+| Project-site deployment under `/agentsmesh` keeps the base path in canonical URLs and sitemap links | ✓ | `website/site-url.test.mjs` |
+| Root custom-domain deployment drops the `/agentsmesh` path entirely | ✓ | `website/site-url.test.mjs`, custom `astro build` |
+| GitHub Pages host does not emit an unnecessary `CNAME` file | ✓ | `website/site-url.test.mjs`, default `astro build` |
+| Custom-domain host emits `CNAME` and root sitemap URL | ✓ | `website/integrations/seo-robots.test.mjs`, custom `astro build` |
+| Built HTML for a root custom domain contains no stale `/agentsmesh` href/src references | ✓ | `rg -n 'href="/agentsmesh/|src="/agentsmesh/|https://docs.agentsmesh.dev/agentsmesh' website/dist -g '*.html'` |
+
+### Gaps Identified
+
+- none in the implemented website changes; the live site still needs `DEPLOY_SITE_URL` pointed at a root custom domain plus DNS hostname redirection for the SEO report to stop seeing the GitHub Pages project-path redirect
+
+### Actions Taken
+
+- proved the current GitHub Pages project-path build already generated `robots.txt`, then fixed the underlying deploy contract so the site can also ship from a root custom domain where root-level SEO files and non-redirected status codes are possible
+- protected the new behavior with unit tests and deploy-workflow coverage instead of leaving it as a one-off config tweak
+
 # TypeScript error repair pass
 
 - [x] Run `pnpm typecheck` and capture the full current TypeScript error set
@@ -218,6 +274,13 @@
 
 | Criterion | Covered by test? | Status |
 | --- | --- | --- |
+
+# Website SEO issue repair pass
+
+- [ ] Reproduce the current website SEO output and confirm which issues are fixable in-repo vs blocked by hosting/domain constraints
+- [ ] Add failing tests first for the website SEO artifacts and URL behavior we can enforce from the repo
+- [ ] Implement the minimal website/deploy changes to fix the reproducible SEO issues
+- [ ] Run verification plus post-feature QA and append review notes
 | Recover the global branch threshold without lowering config | `pnpm test:coverage -- --coverage.reporter=json-summary --coverage.reporter=text-summary` | OK |
 | Add targeted branch tests instead of broad fixture churn | `tests/unit/config/git-remote.test.ts`, `tests/unit/install/install-manifest.test.ts`, `tests/unit/install/git-pin.test.ts`, `tests/unit/install/install-conflicts.branches.test.ts` | OK |
 | Keep full suite stable under coverage load | `tests/unit/cli/commands/watch.test.ts`, full `pnpm test:coverage` run | OK |

website/astro.config.mjs

@@ -2,15 +2,16 @@
 import { defineConfig } from 'astro/config';
 import starlight from '@astrojs/starlight';
 import seoRobotsIntegration from './integrations/seo-robots.mjs';
-import { absoluteFromBase, getSiteOrigin } from './site-url.mjs';
+import { absoluteFromBase, fromBase, getSiteBase, getSiteOrigin, resolveDeploySite } from './site-url.mjs';
 
+const deploySite = resolveDeploySite();
 const site = getSiteOrigin();
 const ogImage = absoluteFromBase('/og-image.png');
 
 export default defineConfig({
   site,
   trailingSlash: 'always',
-  base: '/agentsmesh',
+  base: getSiteBase(),
   integrations: [
     starlight({
       title: 'AgentsMesh',
@@ -52,7 +53,7 @@ export default defineConfig({
         },
         {
           tag: 'link',
-          attrs: { rel: 'icon', href: '/agentsmesh/favicon.svg', type: 'image/svg+xml' },
+          attrs: { rel: 'icon', href: fromBase('/favicon.svg'), type: 'image/svg+xml' },
         },
       ],
       sidebar: [
@@ -123,6 +124,6 @@ export default defineConfig({
         },
       ],
     }),
-    seoRobotsIntegration(() => getSiteOrigin()),
+    seoRobotsIntegration(() => deploySite.publicUrl),
   ],
 });

website/integrations/seo-robots.mjs

@@ -1,20 +1,36 @@
 import { writeFileSync } from 'node:fs';
 
+import { getCnameValue, resolveDeploySite } from '../site-url.mjs';
+
+export function buildRobotsTxt(raw = null) {
+  const { publicUrl } = resolveDeploySite(raw);
+  return `User-agent: *
+Allow: /
+
+Sitemap: ${publicUrl}/sitemap-index.xml
+`;
+}
+
+export function buildSeoArtifacts(raw = null) {
+  const artifacts = [{ fileName: 'robots.txt', content: buildRobotsTxt(raw) }];
+  const cname = getCnameValue(raw);
+  if (cname) {
+    artifacts.push({ fileName: 'CNAME', content: cname });
+  }
+  return artifacts;
+}
+
 /**
- * @param {() => string} getOrigin Host-only HTTPS URL, no trailing slash
+ * @param {() => string | null | undefined} getDeploySiteUrl Full public site URL
  */
-export default function seoRobotsIntegration(getOrigin) {
+export default function seoRobotsIntegration(getDeploySiteUrl) {
   return {
     name: 'seo-robots',
     hooks: {
       'astro:build:done': ({ dir }) => {
-        const origin = getOrigin().replace(/\/$/, '');
-        const body = `User-agent: *
-Allow: /
-
-Sitemap: ${origin}/agentsmesh/sitemap-index.xml
-`;
-        writeFileSync(new URL('robots.txt', dir), body, 'utf8');
+        for (const artifact of buildSeoArtifacts(getDeploySiteUrl())) {
+          writeFileSync(new URL(artifact.fileName, dir), artifact.content, 'utf8');
+        }
       },
     },
   };

website/integrations/seo-robots.test.mjs

@@ -0,0 +1,43 @@
+import test from 'node:test';
+import assert from 'node:assert/strict';
+
+import { buildSeoArtifacts, buildRobotsTxt } from './seo-robots.mjs';
+
+test('buildRobotsTxt points crawlers at the public sitemap URL', () => {
+  assert.equal(
+    buildRobotsTxt('https://samplexbro.github.io/agentsmesh/'),
+    `User-agent: *
+Allow: /
+
+Sitemap: https://samplexbro.github.io/agentsmesh/sitemap-index.xml
+`,
+  );
+});
+
+test('buildSeoArtifacts adds CNAME for custom domains and skips it for github.io', () => {
+  assert.deepEqual(buildSeoArtifacts('https://samplexbro.github.io/agentsmesh/'), [
+    {
+      fileName: 'robots.txt',
+      content: `User-agent: *
+Allow: /
+
+Sitemap: https://samplexbro.github.io/agentsmesh/sitemap-index.xml
+`,
+    },
+  ]);
+
+  assert.deepEqual(buildSeoArtifacts('https://docs.agentsmesh.dev/'), [
+    {
+      fileName: 'robots.txt',
+      content: `User-agent: *
+Allow: /
+
+Sitemap: https://docs.agentsmesh.dev/sitemap-index.xml
+`,
+    },
+    {
+      fileName: 'CNAME',
+      content: 'docs.agentsmesh.dev\n',
+    },
+  ]);
+});

website/package.json

@@ -5,6 +5,7 @@
   "private": true,
   "scripts": {
     "dev": "astro dev",
+    "test": "node --test",
     "build": "astro build",
     "preview": "astro preview",
     "astro": "astro"

website/site-url.mjs

@@ -1,23 +1,62 @@
+const DEFAULT_DEPLOY_SITE_URL = 'https://samplexbro.github.io/agentsmesh/';
+
+function normalizeBasePath(pathname) {
+  const trimmed = pathname.replace(/\/+$/, '');
+  return trimmed === '' ? '/' : trimmed;
+}
+
 /**
- * Single source of truth for the docs site origin. Set DEPLOY_SITE_URL in CI
- * (GitHub repository variable) to your indexed hostname — e.g. apex HTTPS URL
- * with no trailing slash. Configure DNS/CDN to 301 the non-canonical host (www vs apex).
+ * Single source of truth for the docs site's public URL.
+ * Set DEPLOY_SITE_URL in CI to the exact indexed website URL:
+ * - GitHub Pages project site: https://samplexbro.github.io/agentsmesh/
+ * - Custom domain at root: https://docs.agentsmesh.dev/
  */
-
-/** @returns {string} e.g. https://samplexbro.github.io */
-export function getSiteOrigin() {
-  const raw =
+export function resolveDeploySite(raw = null) {
+  const value =
+    raw?.trim() ||
     process.env.DEPLOY_SITE_URL?.trim() ||
     process.env.SITE_URL?.trim() ||
-    'https://samplexbro.github.io';
-  return raw.replace(/\/$/, '');
+    DEFAULT_DEPLOY_SITE_URL;
+  const url = new URL(value);
+  const basePath = normalizeBasePath(url.pathname);
+  const publicUrl = `${url.origin}${basePath === '/' ? '' : basePath}`;
+
+  return {
+    origin: url.origin,
+    basePath,
+    publicUrl,
+    hostname: url.hostname,
+  };
+}
+
+/** @returns {string} e.g. https://samplexbro.github.io */
+export function getSiteOrigin(raw = null) {
+  return resolveDeploySite(raw).origin;
+}
+
+/** @returns {string} e.g. /agentsmesh or / */
+export function getSiteBase(raw = null) {
+  return resolveDeploySite(raw).basePath;
 }
 
 /** @param {string} pathWithLeadingSlash path after base, e.g. /og-image.png */
-export function absoluteFromBase(pathWithLeadingSlash) {
-  const origin = getSiteOrigin();
+export function fromBase(pathWithLeadingSlash, raw = null) {
   const suffix = pathWithLeadingSlash.startsWith('/')
     ? pathWithLeadingSlash
     : `/${pathWithLeadingSlash}`;
-  return `${origin}/agentsmesh${suffix}`;
+  const basePath = getSiteBase(raw);
+  return basePath === '/' ? suffix : `${basePath}${suffix}`;
+}
+
+/** @param {string} pathWithLeadingSlash path after base, e.g. /og-image.png */
+export function absoluteFromBase(pathWithLeadingSlash, raw = null) {
+  const suffix = pathWithLeadingSlash.startsWith('/')
+    ? pathWithLeadingSlash
+    : `/${pathWithLeadingSlash}`;
+  return `${resolveDeploySite(raw).publicUrl}${suffix}`;
+}
+
+export function getCnameValue(raw = null) {
+  const { hostname } = resolveDeploySite(raw);
+  return hostname.endsWith('.github.io') ? null : `${hostname}\n`;
 }

website/site-url.test.mjs

@@ -0,0 +1,45 @@
+import test from 'node:test';
+import assert from 'node:assert/strict';
+
+import {
+  absoluteFromBase,
+  fromBase,
+  getCnameValue,
+  resolveDeploySite,
+} from './site-url.mjs';
+
+test('resolveDeploySite keeps the GitHub Pages project path as base', () => {
+  const deploySite = resolveDeploySite('https://samplexbro.github.io/agentsmesh/');
+
+  assert.deepEqual(deploySite, {
+    origin: 'https://samplexbro.github.io',
+    basePath: '/agentsmesh',
+    publicUrl: 'https://samplexbro.github.io/agentsmesh',
+    hostname: 'samplexbro.github.io',
+  });
+});
+
+test('resolveDeploySite supports a custom domain at the site root', () => {
+  const deploySite = resolveDeploySite('https://docs.agentsmesh.dev/');
+
+  assert.deepEqual(deploySite, {
+    origin: 'https://docs.agentsmesh.dev',
+    basePath: '/',
+    publicUrl: 'https://docs.agentsmesh.dev',
+    hostname: 'docs.agentsmesh.dev',
+  });
+});
+
+test('fromBase and absoluteFromBase honor the resolved base path', () => {
+  assert.equal(fromBase('/favicon.svg', 'https://samplexbro.github.io/agentsmesh/'), '/agentsmesh/favicon.svg');
+  assert.equal(fromBase('/favicon.svg', 'https://docs.agentsmesh.dev/'), '/favicon.svg');
+  assert.equal(
+    absoluteFromBase('/og-image.png', 'https://docs.agentsmesh.dev/'),
+    'https://docs.agentsmesh.dev/og-image.png',
+  );
+});
+
+test('getCnameValue only emits a CNAME record for custom domains', () => {
+  assert.equal(getCnameValue('https://samplexbro.github.io/agentsmesh/'), null);
+  assert.equal(getCnameValue('https://docs.agentsmesh.dev/'), 'docs.agentsmesh.dev\n');
+});

website/src/content/docs/canonical-config/agents.mdx

@@ -60,7 +60,7 @@ Suggest concrete fixes, not vague recommendations.
 
 ## Tool-specific behavior
 
-See the **Agents** row in the [supported tools matrix](/agentsmesh/reference/supported-tools/) for per-target support levels (native, embedded, or unsupported).
+See the **Agents** row in the [supported tools matrix](../reference/supported-tools/) for per-target support levels (native, embedded, or unsupported).
 
 <Aside type="tip">
   Embedded agents carry metadata so that `agentsmesh import` can restore them from their projected form back to canonical `agents/*.md` format without data loss.

website/src/content/docs/canonical-config/commands.mdx

@@ -39,7 +39,7 @@ The filename (without `.md`) becomes the slash command name. The above file at `
 
 ## Tool-specific behavior
 
-See the **Commands** row in the [supported tools matrix](/agentsmesh/reference/supported-tools/) for per-target support levels (native, embedded, or unsupported).
+See the **Commands** row in the [supported tools matrix](../reference/supported-tools/) for per-target support levels (native, embedded, or unsupported).
 
 <Aside type="tip">
   Embedded commands carry AgentsMesh metadata comments so that `agentsmesh import` can restore them to the original `commands/*.md` format without data loss.