Yo Dev! Here’s How I Use AI Tools Like Codex CLI and Aider to Speed Up My Coding
Hey there 👋
If you’re anything like me, you’ve probably been drowning in code at some point, thinking,
“Man, I wish I had a junior dev I could just tell what I want done…”
Well — good news — you kinda do now.
Let me show you how I use Codex CLI and Aider (plus a cool mode called --architect) to make coding way easier, especially if you’re into using AI locally (LM Studio, Ollama, you name it).
🔧 What These Tools Actually Do
Here’s the quick version:
| Tool | What it’s for |
|---|---|
| Codex CLI | You type a prompt in terminal, it analyzes code and gives ideas or code snippets — like “Explain this class” or “Refactor this function” |
| Aider | You chat with your code — and it edits your real files based on what you say. Super clean with Git support |
| Architect Mode (in Aider) | You describe the app you want to build → it gives you file/folder structure and even fills in basic code to get you started |
🧪 Codex CLI — My Quick Code Sidekick
Install it:
pip install codex-cli
Now try:
codex "Explain what the login function does in auth.py"
or:
codex "Generate a test for the payment() function"
It won’t edit your code — it’s more like asking an AI senior dev:
“Hey, can you help me understand or draft something?”
Perfect for:
- Code reviews
- First drafts
- Brainstorming ideas fast
If you use local models, just point it to LM Studio or Ollama by setting endpoint in your ~/.codex/config.yaml.
🤖 Aider — When You Want the Code Actually Changed
Okay this one’s nuts.
Run this in your project folder:
aider main.py
Now chat like this:
“Make all functions in this file async”
“Replace requests with httpx”
“Add error handling to database calls”
It’ll show you a diff. You say yes or no. Done.
And if you’re using Git (you should), it’s super safe and easy to revert anything.
Great for:
- Refactors
- Bug fixing
- Feature updates
🏗️ Aider --architect — Start a Project from Thin Air
Let’s say you’re building something from scratch.
Run:
aider --architect
Then say something like:
“A FastAPI app with login/signup and todo item CRUD.”
Aider will:
- Propose a file layout
- Fill in skeletons for
main.py,models.py, etc. - Ask if you want to edit/approve/add/remove
- Then write the files for you 🎉
This is awesome when you:
- Don’t want to write boilerplate by hand
- Want to try ideas quickly
- Need a jumpstart on a side project
👀 Which One Should You Use?
| If you want to… | Use this |
|---|---|
| Understand or explore code | codex |
| Change real files live | aider |
| Build a project scaffold fast | aider --architect |
🧠 Local Models? Yup, They Work Too
All of these tools can talk to local models if they support OpenAI’s API format (like LM Studio or Ollama).
Just set this in your config:
llm:
provider: openai
endpoint: http://localhost:1234/v1
api_key: sk-no-need
Try with:
qwen1.5-7bdeepseek-coderphi-3-minicodellama-instruct
💡 Real Examples I Use
| Task | Prompt | Tool |
|---|---|---|
| Build new app | “A CLI app that tracks expenses to a CSV file” | aider --architect |
| Cleanup | “Refactor this view to use class-based API” | aider |
| Review code | “What does the class TokenManager do?” | codex |
| Generate test | “Write a pytest for update_item()” | aider |
🔄 Typical Flow I Use
flowchart TD
A["🧠 I get an idea"] --> B["🏗️ aider --architect creates the base"]
B --> C["📂 Project structure + code stubs"]
C --> D["🤖 aider edits & refactors my files"]
D --> E["🧪 codex gives me analysis & extra ideas"]
🎯 Final Thoughts
Honestly, these tools feel like a cheat code for solo devs and small teams.
You get:
- Less boilerplate
- Fewer bugs
- A teammate who never sleeps 😎
If you haven’t tried them yet — go play! You’ll thank yourself later.
🔗 Links to Check Out
- 💬 Codex CLI GitHub
- 🤖 Aider
- 🧠 LM Studio
- 🖥️ OpenWebUI
Get in Touch with us
Related Posts
- The Accounting Software Your Firm Uses Is Built for Your Clients, Not for You
- 2026年本地大模型(Local LLM)硬件选型实用指南
- Choosing Hardware for Local LLMs in 2026: A Practical Sizing Guide
- Why Your Finance Team Spends 40% of Their Week on Work AI Can Now Do
- 用纯开源方案搭建生产级 SOC:Wazuh + DFIR-IRIS + 自研集成层实战记录
- How We Built a Real Security Operations Center With Open-Source Tools
- FarmScript:我们如何从零设计一门农业IoT领域特定语言
- FarmScript: How We Designed a Programming Language for Chanthaburi Durian Farmers
- 智慧农业项目为何止步于试点阶段
- Why Smart Farming Projects Fail Before They Leave the Pilot Stage
- ERP项目为何总是超支、延期,最终令人失望
- ERP Projects: Why They Cost More, Take Longer, and Disappoint More Than Expected
- AI Security in Production: What Enterprise Teams Must Know in 2026
- 弹性无人机蜂群设计:具备安全通信的无领导者容错网状网络
- Designing Resilient Drone Swarms: Leaderless-Tolerant Mesh Networks with Secure Communications
- NumPy广播规则详解:为什么`(3,)`和`(3,1)`行为不同——以及它何时会悄悄给出错误答案
- NumPy Broadcasting Rules: Why `(3,)` and `(3,1)` Behave Differently — and When It Silently Gives Wrong Answers
- 关键基础设施遭受攻击:从乌克兰电网战争看工业IT/OT安全
- Critical Infrastructure Under Fire: What IT/OT Security Teams Can Learn from Ukraine’s Energy Grid
- LM Studio代码开发的系统提示词工程:`temperature`、`context_length`与`stop`词详解













