Prompt Code Validation

[JohnnyIrvin]

Ensure the returned output is actual code

Is your feature request related to a problem? Please describe.
Sometimes, when running code snippets, the returned output is not actual code but plain text (e.g. code with explanations or other remarks). This could be frustrating for users who expect the output to be valid code they can use for further development or when piping to another program.

Describe the solution you'd like
Create a validator that checks if the returned output is valid code. The validator should parse the output and flag it if it's not actual code. This could be done with additional prompts and a possible argument that attempts to check.

Describe alternatives you've considered
The only alternative is to rely on users to manually check if the returned output is actual code. However, this doesn't allow us to easily pipe or chain commands which is the ultimate goal of the program.

[JohnnyIrvin]

I imagine that if we are able to do add this, we would also be able to get rid of the code --

shell-craft/shell_craft/cli/github.py

Line 67 in 7aa95f7

def from_prompt(prompt: str) -> 'GitHubArguments':

[JohnnyIrvin]

Now that GPT supports "json mode" out of the box --- this should be even easier.

We could also consider Pydantic for forcing outputs to conform to a certain schema.