CLAUDE.md

CLAUDE.md is a Markdown file that Claude Code automatically loads as context, letting a team encode project-specific instructions — coding conventions, build and test commands, architecture notes, and behavioral guardrails — that persist across every session. It functions as durable memory for an AI coding agent: instead of re-explaining a codebase in each prompt, you write it down once, commit it, and the agent reads it on every run.

• Auto-loaded context - Discovered at the repository root (and in parent and home directories) and injected into the model’s working context.
• Project conventions - Build/test commands, code style, directory layout, and “do this, never that” rules in plain Markdown.
• Layered and shareable - Committed to the repo so the whole team shares one source of truth, with personal overrides layered on top.
• Import syntax - Can pull in other files so large instruction sets stay modular rather than one sprawling document.

CLAUDE.md is part of a broader shift toward agent-readable configuration files sitting alongside code — a sibling to .clinerules and other per-agent instruction formats. For API teams it is where governance, style guides, and onboarding knowledge get expressed in a form an agent will actually follow, turning tacit conventions into machine-consumable guidance. As more of the software lifecycle runs through coding agents, a well-maintained CLAUDE.md becomes a real governance surface for how those agents design, change, and document APIs.