Add Self-Validation TDD Mode for Incremental Changes

[udirom]

To avoid the problem of integration failures in large projects, I propose adding a self-validation TDD mode to the Roo. This mode would require Roo to automatically write and execute small, targeted tests to verify each incremental change before proceeding.

It should generate and execute unit tests to validate its changes on smaller code portions as it progresses.
Ensure that each incremental change is tested and validated independently to maintain stability.
Stop further code generation if tests fail.

[kostrse]

I doubt a model can write meaningful tests for itself, but automatically running a lint task (or compile / type-check etc.) like Aider does, would be helpful.

Aider runs linter after each change, consumes the linter errors and runs another prompt to fix them.

[natemartinsf]

I've been prompting roo to do this for my project, and it's been fairly successful. I added to the prompt that it follows test-driven-development practices, and it's done a pretty good job following it, and wrote good tests.

[BradKML]

Seconding this as well but it needs to be more comprehensive, from logging errors to learn from, to understanding human requirements.