Pr Feedback Classifier

The pr-feedback-classifier skill automatically analyzes pull request descriptions and comments to categorize feedback into specific, actionable buckets. It helps teams streamline code review workflows by identifying critical issues, suggestions, or general discussion points without manual sorting.

When to use it

• Automate triage: Instantly sort incoming PRs based on whether they contain bugs, feature requests, or documentation updates.
• Prioritize reviews: Highlight critical security vulnerabilities or breaking changes so senior engineers can address them first.
• Reduce noise: Filter out minor stylistic comments from major architectural discussions to keep focus areas distinct.
• Track sentiment: Detect negative feedback patterns early in the review cycle to prevent merge blockers later.

Key capabilities

• Parses raw PR text and comment threads for semantic meaning.
• Assigns predefined labels or tags based on content analysis.
• Integrates directly into Dagster workflows for automated pipeline triggers.
• Supports custom classification schemas tailored to specific team needs.

Example prompts

• "Classify this pull request feedback as 'bug', 'feature', or 'question'."
• "Identify any critical security concerns in the latest PR comments and tag them accordingly."
• "Summarize the top three issues raised in this PR thread and categorize their severity."

Tips & gotchas

Ensure your classification labels align with existing project conventions to avoid confusion during downstream processing. For best results, provide a few labeled examples during initial setup so the model can learn your team's specific terminology and priority rules.