A beginner-first tutorial site for OpenAI Codex Desktop. It is designed for business users who want useful local tools for Windows and macOS, and want to use Codex as an agent for practical tasks.
Live site: https://codex.ifq.ai
This project is not an API reference and not a flashy demo.
It is built around one practical goal:
help non-developers turn a clear business request into a real local desktop app.
The site guides users through a simple path:
- Understand what Codex Desktop can do.
- Generate a stronger prompt with a fill-in-the-blank form.
- Paste that prompt into Codex and let it build step by step.
- Learn from department cases, agent task cases, recipes, and FAQ.
Typical scenarios include finance reconciliation, platform bills, daily reporting, support helpers, HR onboarding, logistics follow-up, carrier scorecards, procurement quoting, packaging demand, marketing recap, live run sheets, legal tracking, admin records, product feedback triage, sales briefs, order exceptions, support issue analysis, product launch checks, replenishment, supplier quote comparison, competitor pricing, member segmentation, coupon risk, live recaps, category reviews, and review insights.
- Auto-detect browser language and enter
/zhor/enon first visit. - Remember explicit user language choice instead of resetting it later.
- Keep prompt output aligned with the current site language.
- Support deep links without locale prefixes such as
/guideor/faq. - Keep the core app-building path focused on local desktop apps, while showing safe agent-task patterns for broader Codex work.
- Explain everything in business language before technical language.
- Produces a ready-to-copy Codex prompt from a lightweight form.
- Includes quick templates, browser-local history, and prompt language switching.
- Explains Codex in plain language from first launch to packaging and sharing.
- Breaks the learning curve into short daily tasks.
- Covers realistic workflows across finance, operations, support, HR, logistics, procurement, marketing, legal, data, admin, and product teams.
- Shows copy-ready prompts for using Codex as an agent on e-commerce operations: sales briefs, order follow-up, support analysis, product launch checks, inventory, procurement, ads, competitor pricing, member segmentation, coupon risk, live recaps, category reviews, and review insights.
- Adds copy-ready scenarios for order merging, product image packaging, shipment exceptions, product labels, marketplace bill checks, campaign calendars, support macros, creator samples, competitor pricing, live timers, listing audits, and inventory transfer planning.
npm install
npm run devThen open http://localhost:3000.
Useful checks during local testing:
- The root path redirects according to browser language.
- Manual language switching is remembered.
- Bare routes such as
/guideredirect to a locale-prefixed route. - The generator can produce a usable prompt with only goal + features filled in.
src/app/[locale] localized pages
src/components shared UI, home modules, generator, cases
src/data cases and prompt template data
src/i18n locale config and dictionaries
public public assets, manifest, llms.txt, agents.md
npm run deployThe site uses Next.js App Router and deploys to Cloudflare Workers via @opennextjs/cloudflare.
If you deploy from the Cloudflare dashboard, use npm run build as the Build command. The deploy step can stay on npx wrangler deploy, because the build now generates the required .open-next artifacts first.
- Next.js 15
- React 19
- TypeScript
- Tailwind CSS 3.4
- Cloudflare Workers
- Path-based i18n routing (
/zh/*,/en/*)
- Teams helping non-developers adopt Codex.
- Builders creating bilingual prompt-learning experiences.
- Developers looking for a beginner-first Next.js content site reference.
MIT License