Draft
Migrate a Working Claude Code Setup (CLAUDE.md, Hooks, and MCP) to Cursor
Task
What you will complete
Migrate a working CLAUDE.md, hooks, and MCP config to equivalent Cursor behavior and re-test the same task.
Success criteria
- The Cursor project contains the instructions, hooks, and MCP servers you chose to move.
- The same task that already worked in Claude Code completes again from Cursor.
- The evidence section can show both results and the config files that were actually used.
What this is not
Not a comparison of Claude Code and Cursor, and not a from-scratch Cursor setup.
Prerequisites
- A Claude Code setup that already completes a known task, including its CLAUDE.md, hooks, and MCP config.
- Cursor, in an environment where you can re-run that same task.
- Permission to copy project instructions and server commands into the Cursor project.
Environment
Not recorded. A verified revision will list the operating system, the tool builds, and the model identifiers that were present for the run.
1. Inventory the working Claude Code setup
From the project that already completes a known task, collect CLAUDE.md, the hooks configuration, and the MCP server entries. Write the task you will re-run in the same words, so the second attempt is the same task and not a similar one.
2. Move the project instructions
Translate the instructions in CLAUDE.md into the Cursor project rules you will actually test. Keep a copy of the source beside the Cursor rule. The verified page will show both files. This draft pastes neither.
3. Move the hooks
List each Claude Code hook you rely on and the Cursor hook you will try in its place. Do not assume the event names match. The verified page will show the hook that fired, or say plainly that no equivalent was found.
4. Move the MCP servers
Point Cursor at the same servers, with the same commands and arguments, and confirm the tools list. Record any path or secret that resolves for Claude Code and fails for Cursor. Do not edit that difference out of the notes.
5. Re-test the same task
Run the original task from Cursor. Keep that result next to the Claude Code result you started from. The evidence section is where both belong, with the config files. It stays empty until the re-test exists.
Evidence it actually ran
This is the shape of the stamp. It has not been earned, and the example date that follows is a placeholder, not a run: Ran on YYYY-MM-DD (Asia/Shanghai) with the pinned versions.
- Command output
- None recorded. A verified revision will quote the output from the run, not a sample written in advance.
- Tests, CI, or Inspector
- Not run for this draft.
- Repository or config
- No link recorded.
Failure modes
Anticipated only. None of these has been checked against a run of this tutorial.
Unchecked
A hook you rely on has no Cursor event you have tested
The migration is incomplete if a required hook is only described and never seen to fire.
Unchecked
The instructions load, but the constraint is not followed
A rule file can be present while the agent ignores the constraint the original CLAUDE.md enforced.
Unchecked
An MCP path or secret does not resolve in Cursor
A server command that works for Claude Code can fail when the working directory or environment differs.
Related
Other tutorials in Migrate.
Switch an OpenAI Agents SDK App from One Model Provider to Another Without Breaking Tools
After switching model/provider, tools and handoffs still work, proven with tests.
Not yet verified
1 min read
Switch an Existing OpenAI Client to Qwen4-Max Without Rewriting It
Point an existing OpenAI Chat Completions, OpenAI Responses, or Anthropic Messages client at qwen4-max on Beijing Model Studio by changing only the key, base URL, and model, with 免费额度用完即停 / Free Quota Only turned on, then confirm one request with a local check.
Not yet verified
15 min read