New feature - Fix the Build

[TechNickAI]

1. When build fails, a github action is triggered.
2. Look at the error
3. Create a plan, ala https://github.com/irgolic/AutoPR, Use an agentic approach to figure out what the problem is
4. Create a pull request
5. If the tests do not pass, try x more times

Will be made easier with a bridge from Octokat library ↔️ LLM, where the LLM constructs the necessary calls to the API based on the latest LLM documentation

[hanselke]

hey, dont you think that this should be part of the 'debug' command feature?

So instead of catching the build fails (which i assume is where we inject tests to make the build fail), I think we should do it internally in-app rather then thru the CI pipeline.

I'm thinking of treating this as a core part of the multi shot approach to get it to fix stuff. It should still be able to respond to github CI failures + ideally formatted github issues.

Would you be able to provide me a failing test for a simple use case? Still trying to grok the stack.