为 Next.js 项目配置搜索引擎优化
检查项目路由与现有元数据,补齐页面 SEO、索引规则、站点地图和结构化数据,并通过构建与页面检查验证结果。
完整提示词
<context>
The App Router has built-in SEO primitives, and most SEO bugs in Next.js projects come from not using them or using them inconsistently: a static `metadata` export or an async `generateMetadata` per route segment; `metadataBase` in the root layout so relative canonical and Open Graph URLs resolve to the production origin (a missing or wrong `metadataBase` produces localhost or preview URLs); `alternates.canonical` per page; title templates in layouts; `app/sitemap.ts` returning entries (with `generateSitemaps` for very large sites); `app/robots.ts`; and `opengraph-image` files or generated images. Metadata from nested layouts merges shallowly, so a child that sets `openGraph` replaces the parent's whole `openGraph` object.
Other common mistakes: canonicals that include query strings or differ in trailing slash from the real URL; preview and staging deployments indexable because robots rules ignore the environment; JSON-LD rendered with unescaped user content (escape `<` when serialising into a script tag); dynamic routes whose `generateMetadata` fetches data twice instead of sharing a cached fetch; and `notFound()` not called for missing records, so missing pages return 200. In recent versions, `params` in `generateMetadata` is a Promise that must be awaited.
</context>
<task>
Implement SEO in this Next.js project.
Scope: {{scope}}
{{#site_url}}Site URL: {{site_url}}{{/site_url}}
{{#indexing_rules}}Do not index: {{indexing_rules}}{{/indexing_rules}}
1. Read `package.json` for the Next.js version and router, the root layout, how routes fetch data, existing metadata, any `next-seo` or hand-written head tags, `next.config` (trailing slash, i18n, redirects), and how environments are told apart. Follow existing patterns.
2. **Root setup:** set `metadataBase` from an environment variable with the production origin as the source of truth; add a title template and default description; set default Open Graph and Twitter fields.
3. **Per route:** add `metadata` or `generateMetadata` with a unique title and description, `alternates.canonical` that matches the URL actually served (no tracking parameters, consistent trailing slash), and per-page Open Graph. For dynamic routes, reuse the page's data fetch through the framework's request caching, and call `notFound()` for missing records.
4. **Indexing controls:** add `robots` metadata (`index: false`) to private or utility routes in scope, and make `app/robots.ts` disallow everything on non-production environments while production allows crawling and lists the sitemap. A Disallow stops crawling, not indexing, so also check whether the host already sends `X-Robots-Tag: noindex` or password-protects preview deployments, and add a noindex header for them if not.
5. **Sitemap:** `app/sitemap.ts` listing canonical, indexable URLs only, with `lastModified` from real content dates; split with `generateSitemaps` if there could be more than 50,000 URLs.
6. **Structured data:** where in scope, render JSON-LD in a `<script type="application/ld+json">` in the page component, serialised with `JSON.stringify` and every `<` replaced with `\u003c` (for example `.replace(/</g, '\\u003c')`), built from the same data the page shows.
7. Remove duplicates: old head tags or a library that now conflicts with the Metadata API, once the replacement is in place.
8. Build the project and inspect the rendered HTML of a sample of routes (start the production server and fetch the pages): one title, one canonical with the production origin, correct robots tags, valid JSON. Fetch `/robots.txt` and `/sitemap.xml`. Run the existing tests, type check and lint.
</task>
<constraints>
- Do not change URLs, redirects or routing as part of this work; flag them instead.
- Never noindex or disallow production routes that are not in the user's indexing rules without asking.
- Do not hard-code the production origin in many files; keep one source of truth.
- Keep structured data true to visible content.
{{> guardrails/scope-discipline}}
{{> guardrails/investigate-before-answering}}
{{> guardrails/verify-before-done}}
</constraints>
<output_format>
## Findings
What existed and what was missing or wrong, in bullets.
## Changes
One line per file: `path`, what changed.
## Verification
Commands run and real results, including a sample of rendered head tags.
## Follow-ups
Items flagged but not changed.
</output_format>可替换内容
- {{scope}}
- {{#site_url}}
- {{site_url}}
- {{/site_url}}
- {{#indexing_rules}}
- {{indexing_rules}}
- {{/indexing_rules}}
- {{> guardrails/scope-discipline}}
- {{> guardrails/investigate-before-answering}}
- {{> guardrails/verify-before-done}}
分类与标签
编程开发、Next.js、页面元数据、站点地图、结构化数据