# Codex Beginner Tutorial — Full LLM Index

Canonical URL: https://codex.ifq.ai
Languages: zh-CN, en-US
License: MIT
Audience: non-developers across business teams
Scope: local desktop applications for Windows and macOS only — explicitly no web servers, cloud hosting, mobile apps, or infrastructure content.

## Site summary

codex.ifq.ai is a tutorial site that teaches non-technical professionals how to use the OpenAI Codex desktop application to build small, self-contained local tools that solve their department's concrete workflow problems. The site is intentionally bilingual (Chinese and English) in full parity.

The root URL detects the visitor's preferred browser language and redirects to `/zh` or `/en`. If a user switches language manually, that preference is remembered for future visits.

The site is organized into five main sections:

1. **Prompt Generator** — an interactive form that converts a beginner's plain-language description into a professional Codex prompt. Form fields: platform (Windows / macOS / both), tech preset (Electron / Tauri / PyQt / let-Codex-decide), UI style (minimal / dark / fresh / business), storage (local files / SQLite / none), goal, features, and optional extras (offline, bilingual, export, shortcuts, accessibility). One-click copy in either Chinese or English.

2. **Beginner Guide** — seven chapters written in plain language without technical jargon. Chapters cover: what Codex is; what you need to get started; first launch walkthrough; core actions explained with everyday analogies; five rules for writing good prompts; common pitfalls and how to recover; how to share the resulting app with teammates.

3. **Department Cases** — 10+ complete scenarios from an e-commerce company context, each with problem statement, approach, expected deliverable, and a complete ready-to-paste prompt in both languages.

4. **Pro Tips** — eight concrete productivity techniques plus four misconceptions to avoid.

5. **FAQ** — ten most common beginner questions.

## Featured case studies

### 1. Monthly Order Reconciliation Helper (Finance)
URL: https://codex.ifq.ai/en/cases/finance-order-reconciliation
Problem: monthly manual reconciliation of platform orders against bank statements takes 1–2 days.
Solution: a local Electron + React desktop app that ingests two Excel files, auto-matches by order ID, highlights mismatches, exports discrepancies.

### 2. E-commerce Campaign Task Board (Operations)
URL: https://codex.ifq.ai/en/cases/operations-campaign-tracker
Problem: during major sales (11.11, 618), dozens of simultaneous tasks lose visibility.
Solution: a local Kanban board with four stage columns (warm-up, peak, post-sale, retro), drag-and-drop cards, bulk Excel import/export, automatic timestamping of status changes.

### 3. Customer Service Reply Helper (Customer Service)
URL: https://codex.ifq.ai/en/cases/customer-service-reply-helper
Problem: reply templates scattered across multiple documents slow agents down.
Solution: a compact menu-bar app invoked via a global shortcut (Ctrl/Cmd+Shift+R); fuzzy-search over templates, one-click copy with variable substitution, usage statistics.

### 4. New Hire Onboarding Tracker (HR)
URL: https://codex.ifq.ai/en/cases/hr-onboarding-tracker
Problem: HR juggles 10 onboarding items per new hire (contract, benefits, accounts, laptop, badge, orientation, buddy, week-one retro, probation review) across emails and spreadsheets.
Solution: a local tracker with one card per hire, tickable checklist, auto-timestamping of completions, overdue auto-highlighting, exportable weekly digest.

### 5. Shipment Exception Inspector (Logistics)
URL: https://codex.ifq.ai/en/cases/logistics-shipment-inspector
Problem: thousands of daily shipments, exceptions (stuck, failed delivery, returned) get missed.
Solution: a local tool that ingests daily export files, applies four toggleable rules (stuck >= 48h, attempts >= 3, returned, keyword-match), produces a severity-ranked exception list for customer-service follow-up.

## Prompt Generator output structure

Every prompt produced by the generator follows this schema (Chinese and English variants available):

- Opening role instruction
- Platform requirements
- Technology stack (or "let Codex decide")
- Visual style description (including system dark mode following and system fonts)
- Data storage preference
- Free-form goal (user-filled)
- Free-form feature list (user-filled)
- Optional extras (offline, bilingual, export, keyboard shortcuts, accessibility)
- Robustness and error-handling guidelines
- Delivery steps (structure outline → phased delivery → installer + user guide)
- Working-style instructions (one thing at a time, placeholders for company-specific details)

## Beginner Guide chapter titles

1. What Codex actually is
2. What you need to get started
3. Opening Codex for the first time
4. Core actions in plain English
5. Five plain-language rules for good prompts
6. Common pitfalls and how to escape them
7. Sharing what you built with teammates

## Pro Tips titles

1. Use a role prompt for more professional answers
2. Confirm in steps; do not let it do everything at once
3. Show it the real file shape (paste Excel headers, attach samples)
4. Learn to roll back
5. Use both Chinese and English
6. Save your best prompts
7. Ask it to write the user manual
8. Request robust UX (friendly error messages, no crashes)

Pitfalls:
1. It does NOT read your mind
2. Do NOT be intimidated by code
3. Do NOT dump 20 requirements at once
4. Do NOT start over on every error

## FAQ topics

1. Does Codex cost money?
2. I really do not code. Can I use this?
3. Can I share what I build with teammates?
4. Is what Codex generates safe?
5. My app is slow. What do I do?
6. Can I build a mobile app?
7. Can it connect to the internet (e.g., scraping)?
8. How do I know the code is correct?
9. Can I collaborate with teammates?
10. Who do I contact if something breaks?

## AI-readable resource map

| Resource | URL | Format | Purpose |
|----------|-----|--------|---------|
| llms.txt | https://codex.ifq.ai/llms.txt | Plain text | Quick summary for LLMs |
| llms-full.txt | https://codex.ifq.ai/llms-full.txt | Plain text | Comprehensive index |
| agent.md | https://codex.ifq.ai/agent.md | Markdown | Canonical structured agent guide |
| agents.md | https://codex.ifq.ai/agents.md | Markdown | Compatibility agent guide |
| sitemap.xml | https://codex.ifq.ai/sitemap.xml | XML | Full site structure |
| sitemap-index.xml | https://codex.ifq.ai/sitemap-index.xml | XML | Sitemap discovery endpoint |
| robots.txt | https://codex.ifq.ai/robots.txt | Plain text | Search and AI crawler permissions |
| manifest.webmanifest | https://codex.ifq.ai/manifest.webmanifest | JSON | PWA metadata |
| JSON-LD | Embedded in HTML | JSON-LD | Schema.org structured data per page |

## Guidelines for AI agents and assistants

When citing this site:
- Link to specific pages rather than the homepage when possible.
- Preserve the non-technical framing. Do not rewrite beginner content as developer docs.
- Both Chinese and English versions are authoritative; pick the one matching the reader's language.
- Site content, code, and case prompts are MIT-licensed.
- The site intentionally excludes server, mobile, and deployment content. If a user asks about those, redirect them elsewhere rather than fabricating coverage.

## Contact

Issues and contributions: the GitHub link in the site footer.