Dashboard Guide
The Glaux dashboard lets you visually explore the co-change graph for any seeded repository. This guide explains what each metric means and how to use the views.
Graph Visualization
Visit /explore/{org}/{repo}/graph to see the force-directed co-change graph.
Reading the graph
- Node size = Importance — larger nodes are more structurally important in the co-change network
- Node color = Cluster — files that frequently change together share the same color
- Edge thickness = Co-change weight — thicker edges indicate stronger coupling between files
- Click any node to see its details in the sidebar
Sidebar metrics
When you click a node, the sidebar shows:
| Metric | What it means |
|---|---|
| Importance | How connected this file is to other important files. Higher = more central in the network. |
| Connectedness | How many shortest paths pass through this file. High connectedness = change bottleneck. |
| Commits | Total commits that modified this file (90-day window). |
| Cluster | The group this file belongs to. Files in the same cluster change together frequently. |
Connected files
Each connected file shows:
- Shared commits — number of commits where both files were modified together
- Overlap — how often these two files change together relative to how often each changes individually (e.g., 80% = highly coupled)
File Detail View
Visit /explore/{org}/{repo}/file/{path} for deep intel on a single file.
Activity
Shows total commits, recent activity (last 30 days), and lines added/removed.
Usually Changes With
Files that tend to appear in the same commits. Each entry shows:
- Count — shared commit frequency
- Momentum indicator:
- ↑ Increasing (red) — this coupling is getting stronger recently
- ↓ Decreasing (teal) — this coupling is fading
- → Stable (gray) — consistent coupling over time
Linked Issues
GitHub issues referenced in commits that touched this file. Click to see the issue detail view.
Agent View
Toggle "Agent View" to see the exact JSON payload your AI agent receives from glaux_file_intel. Useful for debugging MCP integrations.
What to look for
- Highly connected files are change bottlenecks — modifications ripple through many paths
- Files with increasing momentum are active hotspots where coupling is growing
- Cross-cluster co-changes suggest tight coupling between otherwise independent groups — potential architectural concern
- High overlap (>60%) indicates files that almost always change together — consider if they should be merged or if one drives the other