An AI agent that generates docker compose for applications then saves all of them then executes them and finally provide feedback if failure to AI model.

This repository tries to solve docker compose deployment problems through an AI Agent.

Pre Requisites

Installation of docker

Install docker from here

Installation of docker-compose

Install docker compose from here

Now install requirements.txt

Which ever python installation you are running be it virtual environment or anaconda install pip install -r requirements.txt

Construct RAG

For generating RAG you need to clone some existing docker repositories

mkdir -p repos
cd repos
git clone https://github.com/Haxxnet/Compose-Examples.git

git clone https://github.com/docker/awesome-compose.git 

git clone https://github.com/ruanbekker/awesome-docker-compose.git

Select either offline or online model. It could be ollama or online using

openrouter free key.

Note

If you are using github codespaces then it is very easy to try this just use

chmod +x setup.sh
./setup.sh

Model selection

Offline models

Installation of ollama

install ollama from here

Install offline models of deep seek

In one terminal keep running the following

ollama serve 

Also download these models

ollama pull deepseek-r1:1.5b 
ollama pull deepseek-r1:7b
ollama pull deepseek-r1:32b
ollama pull deepcoder:14b 
ollama pull llama3.2:3b 
ollama pull llama3.2:1b
ollama pull deepcoder:1.5b
ollama pull gemma3:4b 
ollama pull gemma3:12b 
ollama pull gemma3:4b

After downloading test the models using by first downloading curl and then use

curl http://localhost:11434/api/generate -d '{
  "model": "deepseek-r1:1.5b",
  "prompt": "one word about me just one word",
  "stream": false
}'

Make sure it works and it keeps on running at 11434

Online model (free tier)

1. Go to [openrouter](https://openrou