Arbor revolutionizes AI-assisted code intelligence by moving beyond traditional embedding-based RAG systems. Instead of treating code as a 'bag of text,' Arbor constructs a detailed, graph-native representation of your entire codebase, mapping functions, classes, and dependencies as nodes and edges. This deterministic program understanding enables AI agents, via the Model Context Protocol (MCP), to navigate the actual call graph, discover logic flows, predict the impact of changes, and retrieve semantically relevant code context. It features incremental persistence, cross-file resolution, and a visualizer to make AI reasoning transparent and inspectable, ensuring precision and trust in large or complex projects.
주요 기능
01Graph-native code representation for deterministic program understanding across multiple programming languages.
02Model Context Protocol (MCP) bridge (ArborQL) for AI agents to query and navigate the code graph.
03Impact analysis and refactoring preview with blast radius simulation before code changes are made.
04Cross-file resolution (World Edges) and global symbol table for accurate dependency tracking across an entire codebase.
0559 GitHub stars
06Logic Forest Visualizer for inspecting AI reasoning, exploring code structure, and visualizing change impact.
사용 사례
01Enhancing AI coding assistants with deterministic program understanding for more accurate context and insights.
02Performing safe, AI-assisted refactoring by predicting the blast radius of changes.
03Navigating and understanding complex, large-scale codebases with visual tools and graph queries.
