Every coding agent has its own word for the same few ideas. Cursor has rules. Copilot has custom instructions. Cline has a rules file and MCP tools. Aider users keep a conventions file. Claude Code has CLAUDE.md, rules, skills, hooks, subagents, MCP servers, plugins, and output styles, which is a lot of nouns for someone who just wants their old setup to work.
Most of it maps over cleanly. Here's where each piece lands, using Anthropic's own guide to extending Claude Code as the reference.
Your rules file becomes CLAUDE.md. Anything you'd put in .cursorrules, .clinerules, .github/copilot-instructions.md, or an Aider CONVENTIONS.md goes in a CLAUDE.md at the root of your repo. It loads at the start of every session. Anthropic's rule of thumb is to keep it under 200 lines.
You may not have to copy anything by hand. Running /init in a repo reads Cursor rules from .cursor/rules/ or .cursorrules and Copilot rules from .github/copilot-instructions.md, according to the memory docs. With CLAUDE_CODE_NEW_INIT=1 set, it also picks up AGENTS.md, Windsurf rules, Devin rules, and .clinerules. If your repo already has an AGENTS.md and no CLAUDE.md, Claude Code version 2.1.277 or later reads it directly with no setup at all.
For a fuller move, the /import command brings configuration from OpenAI Codex, Google Gemini CLI, or Cursor into Claude Code. That covers instruction files, MCP servers, commands, subagents, and skills. Run it with --dry-run first to see what it would write. Cursor imports need version 2.1.265 or later.
Cursor's glob-scoped rules become path-scoped rules. If you had a rule that only applied to *.tsx files, put it in .claude/rules/ with a paths field in the frontmatter. Per the docs, rules with paths only load when Claude works with matching files, so they don't eat context the rest of the time.
Custom slash commands become skills. Anthropic merged the two. The skills docs say a file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy and work the same way. Old command files keep working, but new ones should be skills because skills can carry supporting files like scripts and templates. Claude also loads only a skill's description at session start and the full content when the skill is used, which is why a skill is the place for long procedures you don't need every day.
MCP tools stay MCP tools. MCP is an open protocol, so a server you used with Cline or Cursor usually works in Claude Code with claude mcp add. The MCP docs say only tool names and server instructions load at session start, with full schemas fetched when Claude needs a specific tool, so adding servers costs little context. We covered which ones are worth it in what MCP actually is.
The pieces other tools don't have
Hooks are scripts that run at fixed points, like before a tool runs or after a file is edited. They exist because a written instruction can be ignored. Anthropic's overview puts it bluntly: "An instruction like "never edit .env" in CLAUDE.md or a skill is a request, not a guarantee." A PreToolUse hook that blocks the edit is the guarantee. Hooks cost no context unless they return something.
Subagents are separate Claude instances with their own context window, tools, and permissions. Send one to read 40 files and your main conversation gets a summary instead of 40 files. Our post on when to use a subagent goes into when it's worth it.
Plugins bundle skills, hooks, subagents, and MCP servers into one installable unit, and a marketplace is just a repo with a catalog file listing plugins. You don't need a plugin to use any of those features. The plugin docs also note that an enabled plugin is part of every session, not only the ones where you use it, so don't install ten because they looked useful.
Output styles change how Claude answers, not what it knows. If you liked a terse assistant elsewhere, that's an output style, not a CLAUDE.md line.
Anthropic's advice on order is simple: start with CLAUDE.md, and add the rest when you hit a reason. Its first trigger is "Claude gets a convention or command wrong twice," which means it goes in CLAUDE.md. A long workflow you keep re-explaining becomes a skill. A rule that must hold every time becomes a hook. Most people coming from Cursor need the first one on day one and maybe one of the others by the end of the month.
Sources
Claude Code docs: Extend Claude Code - What each extension type is, what loads into context and when, the 200-line CLAUDE.md rule of thumb, path-scoped rules, the request-versus-guarantee line on hooks, output styles, and the order teams add features.
Claude Code docs: How Claude remembers your project - The rules files /init reads from Cursor, Copilot, Windsurf, Devin, and Cline, and reading AGENTS.md directly in version 2.1.277 or later.
Claude Code docs: Commands - /import bringing configuration from Codex, Gemini CLI, or Cursor, the --dry-run flag, and the version needed for Cursor.
Claude Code docs: Skills - Custom commands merged into skills and the recommendation to prefer skills for new work.
Claude Code docs: MCP - Only tool names and server instructions loading at session start.
Claude Code docs: Plugins - What a plugin bundles, what a marketplace is, and enabled plugins being part of every session.
