Autter provides conversational codebase help and actionable feedback inside the review workflow.
Pull request feedback
When Autter finds a problem, the comment should explain:
- what behavior is affected
- why the behavior is risky or inconsistent
- which repository context supports the finding
- how the author can fix or investigate it
Codebase questions
Use AI chat to ask about complex logic, architecture, repository patterns, or a proposed change. Useful questions include:
- Where is authorization enforced for this route?
- Which services call this module?
- What pattern does this repository use for retries?
- Why does this pull request change the cache invalidation path?
- Which tests cover the failure mode in this finding?
Ask focused questions and name the relevant module or behavior. Specific questions make it easier to verify the answer against code.
Suggested fixes
Autter can propose concrete fixes for review findings. Before applying one:
- Confirm it addresses the reported behavior.
- Check that it follows repository conventions.
- Run the relevant tests and static checks.
- Review the final diff for unrelated changes.
A suggested fix is generated code. Treat it as a proposal, not as a verified patch.
Improve the feedback
- write narrow, testable review rules
- connect relevant issues and documentation
- keep pull requests focused
- use thumbs feedback where available to help calibrate results
- revise rules that repeatedly flag valid code
