Core FunctionalityReview Output
What the Agent Won't Comment On
Things Burg AI intentionally ignores
What the Agent Won't Comment On
Burg AI is designed to maximize signal and minimize noise. It intentionally ignores:
Pure Preferences
Subjective choices where neither option is objectively better:
- Tabs vs spaces
- Single vs double quotes
- Trailing commas
- Blank line placement
Use a linter for these.
Bikeshedding
Trivial discussions that don't impact code quality:
- Variable naming that's "good enough"
- Import ordering
- Brace style
Already-Reviewed Unchanged Code
If a line wasn't changed in the current PR, it won't be commented on — even if it has issues.
This prevents:
- Noise from unrelated code
- Scope creep in reviews
- Blocking PRs for pre-existing issues
Files That Are Auto-Skipped
| File Type | Why Skipped |
|---|---|
| Binary files | Not text, can't be reviewed |
| Lock files | Generated, shouldn't be changed |
| Build output | Generated code |
| Minified files | Not readable |
| Files over size limit | Protects API usage |
Positive Feedback
By default, Burg AI doesn't say "great job!" — it focuses on issues only.
This keeps reviews:
- Actionable
- Fast to read
- Professional