Code Knowledge Graph
The Code Knowledge Graph shows how The AI Platform understands a project’s codebase after repositories or local folders are connected.
When no repository sources or local folders are connected, the page shows an empty state. Add project context from Configuration to enable indexing.
What It Shows
After code context is available, the Code Knowledge Graph can help surface:
- symbols
- JSX or UI structure
- call graph relationships
- code communities or clusters
- patterns found in the project
When the Graph Becomes Available
Indexing has two independent milestones:
- Structural graph ready. The AI Platform extracts symbols and relationships, persists them, and makes structural and text-based queries available.
- Semantic enrichment. A background worker adds meaning-based search vectors in restart-safe batches. The status view reports its progress, and completed batches improve search without blocking structural queries.
Closing the app or restarting indexing does not discard completed semantic batches. A changed structural graph or embedding model starts a new enrichment generation so incompatible vectors are never mixed.
“Ready” refers to the structural graph. Semantic enrichment may still be running after the graph becomes queryable.
When to Use It
Use the Code Knowledge Graph when you want specialists to reason about code structure instead of individual files only. It is especially useful for architecture questions, refactors, and impact analysis.
Next Steps
- Configuration. Add repositories or folders.
- Git Worktrees. Work in isolated branches.
- Chat. Discuss project context in channels.