Codex操作基礎概念！介面、概念、實際案例！手把手帶你了解！原來可以這樣用？

Welcome to join our channel ➤    / @panscischool   What would happen if you fed your spending records to AI? This episode uses the common expense tracking problem as a case study, teaching you step-by-step how to build your own personalized spending analytics advisor using Codex—not only transforming messy spending records into a visual dashboard, but also allowing you to directly ask it questions like "Which credit card is the most cost-effective for my favorite stores?" and "Which channels currently offer discounts, and how can I maximize my spending?" More importantly, this episode will use the "mentoring a new employee" approach to thoroughly understand the operating logic of an AI Agent: from project creation, writing into the job manual, feeding it just the right amount of data, setting permissions, to organizing its memory, assigning skills and add-ons. Mastering this approach will allow you to apply the same principles not only to Codex, but also to Claude and Gemini. // Download Link // https://openai.com/zh-Hant/codex/ // Further Reading // 00:00 Highlights of this Episode 03:05 A Complete Guide to the Codex Three-Column Interface 05:47 Starting to Assign Work: Training an Agent as a Newcomer 06:09 Step 1 | Give Him a Desk: Creating a Project 06:59 Step 2 | Give Him an Onboarding Manual: agents.md / claude.md 08:07 Step 3 | Hand Him the Right Amount of Information: Using @ to Specify Files 09:25 Step 4 | Setting Permissions: Ask Before You Act vs. Let Him Do It 10:31 Step 5 | Organizing Memory: Short-Term vs. Long-Term Memory 12:26 Step 6 | Skills and Connectors 13:58 Quick Review of Six Key Points Editor-in-Chief: Ke Ya-Yun Host: JuJu Script & Testing: Mai Zhi-Cheng Editing: Yang Hao-Cheng