Learning Timeline
Key Insights
Local vs. Cloud Agents
Local Agents work directly on files within your IDE (similar to Cursor or Claude Code), while Cloud Agents (openai.com/codex) can perform tasks asynchronously on the web even when your computer is offline.
Choosing 'Low' vs 'High' Models
While 'High' models are considered 'smarter,' the 'Low' model is actually sufficient for building complex applications faster while consuming fewer tokens.
'Sleep-and-Build' Productivity Tips
You can use Cloud Agents to handle technical or marketing tasks before you go to bed, so your application is ready or further along by the time you wake up.
Prompts
Building a Next.js MVP
Target:
Codex Local Agent
I created a PRD in project.md please build out the MVP using Nex.js
Step by Step
Project Setup and Code Generation with Codex Local Agent
- Open Visual Studio Code (VS Code) on your computer.
- Click on the 'Codex' icon in the left sidebar to open the Codex plugin interface.
- Click 'New File' or use a shortcut to create a new file in your project.
- Save the file as 'project.md'.
- Copy and paste your entire Product Requirements Document (PRD) into the 'project.md' file.
- In the Codex panel on the left, select the 'Work Locally' option. This allows the AI to see and edit code directly within your IDE.
- Type your prompt in the chat area to instruct the AI to build the application based on the PRD file.
- Select the 'Low' model setting (instead of 'High') for faster generation and to save on tokens.
- Click the 'Send' button to automatically start the Next.js MVP building process.
