How Trevyn Works

System overview

Trevyn combines repository context, PR diffs, and your configured rules to produce actionable review comments. Outcomes are measured through suggestion adoption, review completion, and repository trends.

Pull Request Event
      |
      v
Diff + Repository Context
      |
      v
Rule Evaluation
      |
      v
Review Suggestions
      |
      v
User Feedback -> Learnings

1. Pull request review lifecycle

1. A pull request event triggers a Trevyn review run.
2. Trevyn analyzes changed files and repository context.
3. Suggestions are generated and posted to the review timeline.
4. Applied suggestions are tracked for reporting and usage metrics.

The most successful teams start with smaller repositories, validate suggestion quality, then expand activation gradually.

2. Rules and scope

• Global Rules apply across all repositories.
• Repository Rules apply only to a selected repository.
• Optional file patterns let you target specific code areas, such as backend routes or frontend components.

For predictable results, keep rules explicit, measurable, and technology-aware instead of generic style guidance.

3. Learnings feedback loop

• When you reply to review comments, Trevyn can extract learnings.
• Each learning stores the source pull request link.
• Active learnings influence future review quality.

This creates a team memory system where repeated feedback becomes reusable review guidance.

4. Visibility and outcomes

• Dashboard shows key stats like reviews, time saved, and plan use.
• Analytics shows activity trends and top reviewed repositories.
• Reviews page shows suggestion details and applied counts.

Use dashboard for daily signals and analytics for weekly and monthly planning.

5. Usage limits

Trevyn tracks reviews used during the current monthly period. Free accounts include 25 PR reviews each month, and you can request more credits when you need extra review capacity.

Operational best practices

• Run weekly rule review with one technical owner.
• Compare acceptance rate before and after major rule updates.
• Pause noisy repositories while tuning rule precision.
• Audit learnings monthly to remove stale guidance.