Show HN: Arbor – code graph MCP server so agents stop grep-reading your codebase
Arbor v2.4.0 is a code graph MCP server that helps AI agents navigate codebases structurally instead of using grep. It provides a semantic understanding of code relationships for more efficient agent-based code analysis.
Background
- Arbor is an open-source "code graph" tool that maps how functions, classes, and variables connect across a codebase. Instead of grepping raw files, an AI coding agent can query this graph to find relevant code faster and with richer context.
- MCP (Model Context Protocol) is an open standard from Anthropic that lets AI models talk to external tools. An "MCP server" exposes Arbor's code graph as a service the model can query directly during a coding task.
- v2.4.0 adds Rust support (via tree-sitter parsing), an "affected code" query that finds code changed between git commits, and a "neighbors" query that returns nearby lines around a symbol — making it easier for agents to navigate code without reading entire files.
- The project addresses a real pain: AI coding assistants often struggle with large codebases because they rely on simple text search, missing structural relationships that a human developer would see.