Big Indexer

Architecture-aware MCP context for large codebases — behavioral boundaries, coupling seams, impact scope, and BGI-TWIN guidance.

Big Indexer (BGI) is a static architecture analysis MCP server that groups code by behavioral role using COV tokens rather than file imports. It classifies code units, creates behavioral edges with scoped matching rules, clusters related units with hard size caps, and emits architectural boundaries as machine-readable artifacts. Supports 30+ programming languages across three tiers with incremental scanning, diff analysis, and BGI-TWIN guidance for AI agents navigating large codebases.