gopeed-website-fuma
This is a Next.js application generated with
Create Fumadocs.
Run development server:
npm run dev
# or
pnpm dev
# or
yarn dev
Open http://localhost:3000 with your browser to see the result.
API reference and AI page actions
Use Node.js 24+ and pnpm 9.15.9. pnpm docs:generate downloads the Gopeed
OpenAPI schema and generates the reference under content/api-reference/.
These directories are disposable build output: edit the upstream schema or
scripts/generate-openapi.mjs, rather than generated MDX. Development, type
checks and builds generate the reference automatically.
The reference is an independent section at /api-reference, with its own content
collection, layout and sidebar, and localized equivalents under /zh, /zh-TW
and /ru. Old /docs/openapi URLs redirect to this section. Endpoint names and
descriptions remain in English; navigation,
page actions and the API playground use the selected UI language. The schema is
also available at /openapi.json. Playground requests go directly from the
browser to the selected Gopeed server.
Every docs page offers native Copy Markdown and Open actions. Append .md to
a document URL (use /docs/index.md for the docs home); existing .mdx URLs
remain supported. /llms.txt and /llms-full.txt serve the English index and
full text, with localized equivalents such as /zh/llms.txt. The shared index
and full text include both guides and the independent API reference. Page actions
use the selected language and canonical Markdown URLs. API Markdown
contains a standalone schema fragment including referenced components, rather
than an unrendered React component. No model credentials are needed for these
page actions and text endpoints. These integrations do not include a hosted
Ask AI chat or an MCP server.
SEO URL and sitemap behavior
The sitemap is generated at request time from the current D1 catalog. New
extensions do not require a website rebuild to appear. Successful responses allow
up to five minutes of shared caching; D1 failures return an uncached 503 with a
retry interval instead of a partial sitemap. Store listings read the current
catalog at request time. Canonical URLs, internal links
and sitemap entries consistently encode extension IDs (author%40name).
SEO rollout and monitoring
The current implementation covers language routing and metadata, redirects for
legacy duplicated locale paths, sitemap freshness, crawlable homepage/store
links, shared image crawling and first-screen image loading. Homepage platform
cards link to the corresponding existing extension pages in each language.
Extension detail-page copy and READMEs are localized in English, Simplified
Chinese, Traditional Chinese and Russian. Additional languages and tutorials
remain separate content work.
After deployment:
- Check live canonical/hreflang, legacy redirects and sitemap completeness in
Search Console. Inspect representative English/Chinese home, docs, store and
extension URLs; monitor old duplicated-path 404s over subsequent crawls.
- Compare 28-day Search Console periods by landing page, country and non-brand
query cluster. Review clicks, impressions, click-through rate and average
position without treating impressions as market search volume or installs.
- Review mobile LCP/INP/CLS using field data. The smaller preloaded screenshot
and immediately visible first-screen content address concrete loading delays;
a passing field assessment still requires post-deployment measurements.
- Review the generated extension pages against their source capabilities.
Research English demand in India/US and native-language demand in
Indonesia/Brazil, checking actual feature support and conversion before
expanding languages. Add distinct tutorials for demonstrated user problems.
- Review content accuracy, broken links, index coverage, acquisition and relevant
referring links monthly. Keep an owner and a verified feature/version record
for each priority extension page.
Explore
In the project, you can see:
lib/source.ts: Code for content source adapter, loader() provides the interface to access your content.
lib/layout.shared.tsx: Shared options for layouts, optional but preferred to keep.
| Route |
Description |
app/(home) |
The route group for your landing page and other pages. |
app/docs |
The documentation layout and pages. |
app/api/search/route.ts |
The Route Handler for search. |
Fumadocs MDX
A source.config.ts config file has been included, you can customise different options like frontmatter schema.
Read the Introduction for further details.
Learn More
To learn more about Next.js and Fumadocs, take a look at the following
resources:
Cross-origin public API access
The public extension list, detail, install-report, and /api/releases endpoints support
credential-free cross-origin requests. worker.ts applies CORS outside the
response cache, including cached results, API errors, and JSON POST preflight.
The authenticated sync endpoint keeps its existing behavior.
Use pnpm cf:dev for browser integration tests through the Cloudflare Worker.
Extension content generation
Apply pnpm db:migrate before deployment and pnpm db:migrate:dev for local
D1 development. Extension source content stays in extensions;
extension_localizations stores the title, description and complete README for
each language, together with its source hash and update time.
The hourly cron requires the existing GITHUB_TOKEN Worker secret and the AI
binding. SEO copy and translations use
@cf/deepseek-ai/deepseek-v4.1-flash-1017 through env.AI.run() with reasoning
disabled. No Cloudflare API token secret is needed for inference.
The cron checks the manifest version first. On a version change, it hashes the
title, description and recognized README files from one commit. Only changed or
missing content triggers generation. Missing languages are retried against the
saved source snapshot even without another version change. Author-provided
translations and original-language READMEs are reused.
lib/i18n/config.ts is shared by website routing, extension generation, response
schemas and missing-language checks. Adding a website
language makes the next cron fill only its missing content, even if extension
versions stay unchanged. Database locale values are not restricted to a fixed
list. Until generation succeeds, the new language's pages show upstream content.
Translation preserves code, URLs and Markdown structure. Chinese interface
labels retain their original text with a model-generated translated explanation.
Long READMEs are translated in sections and validated as a complete document.
Requests are paced below 20/minute, have a 180-second deadline and get one retry;
rate-limit or capacity errors wait one minute before retrying.
Failed generation never overwrites valid content. Pages use translations only
when their source hash matches the current extension. Otherwise, titles,
descriptions and READMEs fall back to the upstream source. Page requests never
call AI. The public extension API continues to return upstream content.