ck:gkg

Semantic code analysis finds cross-file symbol usages by applying AST-based extraction to parse the repository, indexing the relationships in a graph database, and querying those definitions and call sites via HTTP or MCP tools.

Impact analysis before refactoring is best performed by querying a semantic knowledge graph to review all symbol call sites and dependencies, ensuring you understand the full scope of changes before modifying the code.

To index your repository for go-to-definition and code navigation, you need to start the GKG server workflow, which builds the AST-based knowledge graph using KuzuDB so you can query symbol relationships for a specific project path.

Semantic code analysis works better than text search for locating symbol definitions because it understands the code structure rather than relying on brittle string matching, allowing it to accurately find what a symbol refers to across files.

Yes, you can use a knowledge graph to support architecture visualization by discovering and extracting the relationships between components, which helps explain the system structure to both humans and AI during diagram preparation.

A limitation of using AST parsing for semantic code analysis is that it relies on proper indexing via the GKG CLI and requires supported languages, meaning unsupported syntax or unindexed files will not be captured in the graph queries.