Kody Learning

​

Quick Overview

• 👍 Thumbs up: Kody learns what your team likes
• 👎 Thumbs down: Kody learns what to avoid
• ✨ Auto-learning: Kody tracks which suggestions get implemented

​

How the Learning Works

​

1. Collecting Feedback

Kody learns from three main sources:

• Direct Reactions: Team members can react to Kody’s suggestions on GitHub/GitLab:

  • 👍 = “This is what we want”
  • 👎 = “This doesn’t match our style”

• Implementation Tracking: When your team implements Kody’s suggestions, she marks it as positive feedback

• Pattern Recognition: Kody builds knowledge clusters based on your team’s preferences

​

2. Smart Filtering

Once Kody has enough data, she:

1. Analyzes new code suggestions
2. Compares them against your team’s preference clusters
3. Only shows suggestions that match your team’s style
4. Filters out suggestions similar to previously rejected ones

​

Why Your Feedback Matters

The more feedback you provide, the better Kody becomes at:

• ✅ Matching your coding style
• 🎯 Making relevant suggestions
• 🚫 Avoiding unwanted patterns
• 💡 Understanding team preferences

​

Best Practices

1. Be Consistent: Encourage the team to provide feedback regularly
2. React Promptly: Give feedback while the context is fresh
3. Team Alignment: Ensure the team agrees on what makes a good suggestion

​

Coming Soon

• 🔄 Bitbucket support
• 📊 Feedback analytics dashboard
• 🎛️ Custom learning preferences