@byCanen: Handling a client's website cloning request, someone shared a case: a 5,000 yuan order, completed and delivered in 3 ho…

Handling a client’s website cloning request, someone shared a case: a 5,000 yuan order, completed and delivered in 3 hours using REA + Codex.

He said it would have taken two days with the old method. This time’s division of labor was:

REA handles researching the reference site’s page structure, frontend scripts, interactions, and network requests.

Codex handles implementing the functions based on the breakdown results.

Requirement boundaries, testing, and final delivery are still overseen by oneself.

REA was introduced in the previous round; this time, I think this order-taking process can be directly copied.

Installation: npx rea-agents setup

After getting the client’s reference site, you can first give Codex this prompt:

“First, don’t write any code. Use REA for read-only analysis of this reference website that has permission to research: [URL].

Please output:

1. Pages and main functions visible to the user
2. Jumps between pages and core interactions
3. Requests and data structures observable through legitimate access
4. Screenshots, scripts, or request evidence corresponding to each judgment
5. Minimal function list that needs to be implemented for this requirement
6. Content that needs customer confirmation and parts that won’t be done

Do not bypass login, payment, or access controls; do not copy proprietary materials.

After I confirm the scope, then have Codex implement and test.“

This way, before quoting and starting work, you can clarify exactly what needs to be done with the reference site.

govin.eth | G哥 (@goan999999): Codex + REA ，让我接了个 5000 元的订单！

客户甩来一个参考网站：“我也想做个类似的”

这个工具真是万物皆可逆向！REA 帮我研究页面、脚本和请求，Codex 负责写代码，我负责需求、测试和交付

要不是因为这个工具，这点活我要干2天，我3小时干完

你还在收藏 AI 工具，客户已经带着需求来问价了