crg-integration

AST-based context retrieval minimizes token usage by extracting only relevant AST nodes and dependency trees instead of loading entire files. This approach provides AI coding agents with precise semantic context for refactoring while keeping context window consumption low.

AST-based impact analysis generates dependency trees to visualize how code changes affect connected modules. By querying the abstract syntax tree, it maps caller and definition relationships to identify propagation paths without scanning entire files.

Query an AST knowledge graph with a file path and line number to retrieve only the necessary code excerpts and AST nodes. This minimal context injection provides AI agents with the exact semantic information needed without loading entire files.

Yes, the code-review-graph MCP server must be installed and configured within your agent environment. This server provides the AST-based knowledge graph infrastructure required for dependency analysis, semantic code searching, and minimal context retrieval.

Switch to AST-based context retrieval when context window usage exceeds 60 percent. At this threshold, loading entire files becomes inefficient, and injecting only the critical dependency graph and minimal code nodes maintains agent performance for large-scale refactoring tasks.

Yes, AST-level semantic search performs precise queries to find callers or definitions without full-text scanning. By leveraging the knowledge graph, it locates specific code relationships and dependencies directly through abstract syntax tree nodes.