NEW AI Coding Assistant 🤖 Use Prompts To Build New Or EXISTING Apps! (Aider, The Holy Grail)

Aider Github –
Universal CTags –
OpenAI API Keys –
Install Bash on Windows by installing a GIT Client –
Install Conda in Ubuntu WSL2 – [Setting up Jupyter using Conda in WSL2]

Homebrew –

Install Homebrew on Ubuntu WSL2:

/bin/bash -c "$(curl -fsSL"

Install Homebrew in the Anaconda terminal.

bash -c "$(curl -fsSL"

Reopen the terminal to have Homebrew in your path.

If Homebrew is not in your path, add Homebrew to your path:

(echo; echo 'eval "$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)"') >> /home/$USER/.bashrc

In the Windows Anaconda (miniconda) terminal, open bash. (The bash version that opens needs to be the Linux bash version from WSL2)

Fix Conda on the Ubuntu WSL2.
source ~/anaconda3/etc/profile.d/

Use Ubuntu WSL2 or open the Anaconda terminal: (careful with the –name dashes, copy/paste can paste the wrong ascii dash so just type it)
conda create –name aider python=3.11.3
conda activate aider
export OPENAI_API_KEY=12345
brew install universal-ctags

Check if ctags is installed:
ctags --version

Install Aider:
pip install aider-chat

Use Aider to create a Snake game.
mkdir snake
cd snake
create the game snake using python

To get Aider to actually write the changes to I needed another prompt.
save the changes to

In another terminal Window open VS Code to run the generated Python.
cd snake
code .

AI was able to slow down the speed of the snake.
can you slow down the movement speed by half

AI was able to add early quiting the game.
I want to be able to quit while in game by pressing q

AI had trouble increasing the tile size and stopped being able to pick up food.
Can you make the snake game use larger tiles?

I would recommend commit/push the files every time the AI makes a change and it works. That way you can revert the change if the AI breaks something.

Ollama is the single easiest way to run AI models on your laptop


[Github Ollama Project]

[Python Ollama Project]

pip install ollama

Add your local Python scripts folder to your path:



Models are downloaded to %userprofile%.ollama\models.

ollama search

Found 6 available models:

ollama run wizard_vicuna_13b_uncensored

You can change Ollama to not print the prompt.


def generate(*args, **kwargs):
    if prompt := kwargs.get("prompt"):
        #print(">>>", prompt, flush=True)
        generate_oneshot(*args, **kwargs)

    if sys.stdin.isatty():
        return generate_interactive(*args, **kwargs)

    return generate_batch(*args, **kwargs)