Back to Resources
New Laptop Migration GuideLast Updated: Published July 16, 2026

Transfer Claude Code Settings to a New Computer

To transfer Claude Code settings to a new computer, move your personal skills, project skills, supporting files, and workflow notes as complete folders, then verify them with real tasks. The goal is not just copying configuration; it is preserving the AI setup you rely on every day.

Audit
Move
Verify

A practical migration checklist for moving Claude Code skills and reusable AI workflows to a new laptop without losing local-only instructions or rebuilding from memory.

The New Laptop Problem for Claude Code

New laptop setup usually starts with obvious tools: package managers, Git, editors, SSH keys, and a few favorite extensions. Claude Code settings are easier to miss because the best parts of your setup often live as quiet local files: a review skill, a release checklist, a writing style guide, or a repo-specific workflow you built over months.

The mistake is assuming your AI workflow lives in your account. Some parts may, but local skills and project instructions can be machine-bound or repo-bound. If you wipe the old laptop too early, the workflow does not fail loudly; it simply disappears the next time you need it.

A Claude Code migration is successful when the new machine can run the same real workflows as the old one without you rewriting instructions from memory.

What to Transfer Before You Retire the Old Machine

Search intent for transfer Claude Code settings new computer is informational with a setup task attached: people want to know what files and workflows they need before the old laptop is gone. Start with an inventory, not a copy command.

ItemWhy it mattersMigration rule
Personal skillsThey carry your repeatable operating style across projects.Move complete skill folders, not only SKILL.md.
Project skillsThey encode repo-specific build, test, review, or domain conventions.Keep them with the repo unless they are truly reusable everywhere.
Supporting filesExamples, templates, references, and scripts often make the skill useful.Copy the whole folder tree and verify file counts.
Prompt snippetsOld prompts may be the raw material for future skills.Save the best ones, then convert repeatable prompts into skills.

Step 1: Make a Dated Backup First

Before you rearrange anything, create a dated archive from the old machine. This gives you a rollback point if a folder is missed, a cloud sync creates a conflict, or a project checkout excludes hidden files.

# On the old computer
mkdir -p ~/Desktop/claude-code-migration
tar -czf ~/Desktop/claude-code-migration/claude-skills-2026-07-16.tgz -C ~/.claude skills

# Confirm the archive contains complete skill folders
tar -tzf ~/Desktop/claude-code-migration/claude-skills-2026-07-16.tgz | head

If your setup uses project-level skills, back up those folders from each important repo too. Personal skills and project skills solve different problems; the Claude Code skills folder guide explains how to separate them before you move.

Step 2: Move Personal Skills to the New Computer

Personal skills are the ones you expect to use everywhere: code review habits, debugging checklists, writing preferences, research workflows, release notes, or planning templates. These are the highest priority because they define how Claude Code feels on your machine.

# On the new computer
mkdir -p ~/.claude
tar -xzf ~/Downloads/claude-skills-2026-07-16.tgz -C ~/.claude

# Quick inspection
find ~/.claude/skills -maxdepth 2 -name SKILL.md -print

A copied skill is not verified until it works in context. After the folders land, run the same prompt you used on the old computer and compare whether Claude Code finds the right skill and produces the expected structure.

Step 3: Keep Project Skills Attached to Projects

Project skills should move with the repository, not with your global personal setup. If a skill says how to run one app's test suite, reviews one product's domain rules, or references a private architecture convention, it belongs near that project.

The easiest mistake is turning every useful project skill into a global personal skill. That creates context leakage: Claude may apply a repo-specific release checklist in the wrong workspace. Promote only the general pattern, not the private details.

When you do want a project habit everywhere, create a clean general version. For example, turn "review Prompttly Mac release notes against this repo's packaging script" into "review release notes for missing install, rollback, and verification details." Then save the generalized skill separately.

Step 4: Decide What Should Become Portable

A new computer migration is the right moment to stop treating skills as scattered local files. If a workflow matters on one laptop, it probably matters on the next laptop, in another repo, or inside another AI tool.

Prompttly is a skill manager for AI agents — one library for your skills and prompts that syncs into Claude Code, Codex, ChatGPT, and Claude, so your setup follows you across every machine, repo, and tool.

Use the migration to sort your library into three buckets: keep local, make portable, or retire. The goal is not to preserve clutter. It is to keep the workflows that still earn their place.

  • Keep local: One-off experiments, machine-specific notes, and private scratch instructions.
  • Make portable: Repeated workflows you expect to use across laptops, repos, Codex, ChatGPT, Claude, or MCP-connected agents.
  • Retire: Duplicate skills, stale prompts, and old instructions that no longer match how you work.

If you have raw prompts that should become durable workflows, use the free Claude Skill Creator after the migration.

Step 5: Verify the New Machine Before You Wipe the Old One

Do not mark the migration done because the folders exist. Mark it done when the workflows behave correctly.

  1. Confirm every important skill folder exists on the new computer.
  2. Open each folder and confirm the entrypoint is named SKILL.md.
  3. Check that examples, templates, scripts, and references moved with the folder.
  4. Run one realistic task for each high-value skill.
  5. Compare the output against the old machine while it is still available.
  6. Remove duplicates only after the new copy is proven.
  7. Document the final source of truth so the next migration is easier.

If the new laptop will stay in rotation with the old computer, switch from one-time migration to ongoing sync. The Claude Skills sync guide covers that pattern in more detail.

Common New-Laptop Mistakes

  • Copying only visible files: Hidden folders can hold the workflow files you actually need.
  • Moving SKILL.md without its folder: Multi-file skills lose examples, templates, and references this way.
  • Promoting repo-specific rules globally: Project-only context can make Claude Code apply the wrong assumptions elsewhere.
  • Deleting the old laptop copy too soon: Keep it until the new machine passes real workflow tests.
  • Skipping version history: If a migrated skill gets worse, you need an easy way back.

The best migration leaves you with fewer, clearer skills than you started with.

Start from the resources hub for the full Claude Skills series. Use how to install and use Claude Skills for invocation checks, SKILL.md explained for file structure, and how to create a Claude Skill when you need to rebuild an old prompt as a reusable workflow.

People Also Ask About Moving Claude Code to a New Computer

How do I transfer Claude Code settings to a new computer?

Back up your Claude configuration folders, copy personal skills and project skills to the matching locations on the new computer, then verify that each SKILL.md file and supporting file made the move. Test with one real workflow before deleting the old machine copy.

Do Claude Skills automatically move to a new laptop?

No. Skills stored only in local Claude Code folders stay on that machine unless you copy, sync, or manage them in a portable library.

Should I copy personal skills or project skills first?

Copy personal skills first if you are rebuilding your everyday Claude Code setup. Copy project skills with the repo they belong to, because repo-specific instructions should stay attached to that project context.

What is the safest way to avoid losing Claude Skills during migration?

Create a dated backup archive, confirm the archive contains every skill folder, move complete folders instead of individual Markdown files, and keep the old machine untouched until the new machine passes a verification checklist.

Make the next laptop migration boring

Prompttly keeps one versioned skill library and syncs working skills into Claude Code and Codex as real folders, with the same workflows available in ChatGPT, Claude, and MCP-connected agents.