Does anyone know of an AI tool to take a shell script made for setting up an environment in i.e. Ubuntu Server 22.04 and Dockerizing it effectively? (bonus points if it can determine when to grab alpine linux > ubuntu)

Background info:

Right now I have a shell script that installs and configures: Apache or Nginx based on arg1 MySQL (Dynamic schema, table, users/passwords based on arg2) Gets and installs newest LTS version of node (because for some reason Ubuntu is still shipping with 12!)* Gets and installs Laravel, installing with jetstream, git repo, etc.

If I attempt to Dockerize with gpt-4, I'm missing communication between the application and the web server. I assume because it's a non-issue when on the same server/in the bash script, but becomes relevant in the Docker environment (separate containers).

P.S. I've manually fixed it with relative ease - but looking for a robust solution to handle this without manual corrections.

P.S.S. gpt-4 with good system/function messages and solid prompting returns surprisingly good results for a generalized LLM, but I'm looking for an application or library with access to a fine-tuned generalized model, or a model trained specifically for this task.

  • "looking for a robust solution to handle this without manual corrections" --> I don't think there's any AI tool fulfilling this criteria Oct 21, 2023 at 15:03
  • @FliegendeWurst Fair enough, maybe I should say a specific model for dockerization that requires fewer manual corrections. Although I believe it's a valid goal to aim for, even if just by combining a general model with iterative testing until the solution is 100% through automated testing and filtering that trial and error dataset to fine tune a model, hopefully leading to fewer iterations on each test case.
    – TCooper
    Dec 1, 2023 at 14:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.