Problem
When solving an issue, contributors may focus on resolving its functional aspects without taking the project's architectural context into account.
This can lead to subtle problems that do not cause immediate failures, but may affect the maintainability, consistency, or stability of the project.
Examples of these problems include:
- Introducing direct dependencies between layers that should remain decoupled.
- Violating architectural or development guidelines described in the documentation.
- Indirectly affecting other related components.
Keeping all these architectural concepts and documentation details in mind represents a considerable cognitive load, which is even greater for new contributors because they are less familiar with the project's architectural context.
Proposed solution
I suggest exploring the idea of a code review assistant that uses the project's existing documentation to analyze local changes before they reach a human review or even before a pull request is opened.
My idea is to provide it with:
- The issue URL.
- The current git diff.
- Relevant technical or architectural documentation.
Based on that, the assistant could:
- Warn about potential architectural or layering issues.
- Detect regression risks.
- Highlight possible indirect effects on other components.
- Recommend tests related to the implemented change.
The intention is not to automate the code review process, but rather to perform a superficial analysis of the architectural aspects of the code in order to provide early guidance to contributors, with the goal of detecting and fixing problems before a human review or even before a pull request is opened.
I am interested in knowing whether something similar has already been explored before, or whether you think this could be useful in practice.
Problem
When solving an issue, contributors may focus on resolving its functional aspects without taking the project's architectural context into account.
This can lead to subtle problems that do not cause immediate failures, but may affect the maintainability, consistency, or stability of the project.
Examples of these problems include:
Keeping all these architectural concepts and documentation details in mind represents a considerable cognitive load, which is even greater for new contributors because they are less familiar with the project's architectural context.
Proposed solution
I suggest exploring the idea of a code review assistant that uses the project's existing documentation to analyze local changes before they reach a human review or even before a pull request is opened.
My idea is to provide it with:
Based on that, the assistant could:
The intention is not to automate the code review process, but rather to perform a superficial analysis of the architectural aspects of the code in order to provide early guidance to contributors, with the goal of detecting and fixing problems before a human review or even before a pull request is opened.
I am interested in knowing whether something similar has already been explored before, or whether you think this could be useful in practice.