How Claude Code works in large codebases: Best practices and where to start
Summary
The article explains how Claude Code operates in very large codebases, emphasizing live-context navigation over centralized indexing. It outlines the harness architecture (CLAUDE.md, hooks, skills, plugins, MCP servers) plus LSP integrations and subagents, and provides patterns for scalable deployment. It also discusses organizational ownership and governance needed to maximize adoption across large enterprises.