Getting started

Clone-Me remembers the information and judgment standards that you confirm. When an AI agent (such as Claude, Claude Code or Codex) reaches a choice that depends on your taste or standards, Clone-Me gives it a reference judgment based on those standards. If you are new, follow these three steps in order.

1. 1Sign up
2. 2Connect an AI agent
3. 3Use it and review

Step 1. Sign up

Clone-Me is in beta, so only invited people can sign up.

1. Select Sign up in your invitation email

   To enter the invitation code and email yourself, open the sign-up page. You can sign up only with the invited email.

2. Read and check the agreements

   These are the beta terms, the collection and use of personal information, and the overseas transfer consent. Nothing is checked at first.

3. Enter the 6-digit code from your email

   When sign-up is complete, the console's Get started page opens.

Step 2. Connect an AI agent

A connection is one AI agent that calls Clone-Me, for example "Claude Code on my work laptop". In the console, go to Connections → Add connection and choose where you use Clone-Me. The page then guides you step by step for that agent.

• Claude app

  claude.ai on the web, Claude Desktop, Cowork, mobile

  Sign in, no token
• Claude Code

  Claude Code in the terminal, VS Code and JetBrains

  One command + paste a token
• Codex

  Codex CLI, IDE extension, desktop app

  One command + paste a token

• If you are not sure, connect only the one you use most. You can add the others later.
• Other MCP clients that can send an Authorization header can also connect. See the last section of Connecting Claude Code.
• A ChatGPT connector is in preparation.

When you finish, the Check the connection box changes to "Connected" by itself. If it does not change, see Troubleshooting.

Step 3. Use it and review

After you connect, work with your AI agent as usual.

1. First, send your agent "Look up my information in Clone-Me." When the agent calls Clone-Me, the call shows in Call history in the console.
2. When you choose one of several options, correct the AI's suggestion, or give a reason, the agent sends that moment as a case.
3. The server gathers cases from many conversations and writes only repeated or important ways of judging as memory proposals.
4. In the console's Inbox, read each proposal and accept, edit or reject it.

Read next

• How Clone-Me works

  How memories form, dreaming, and the three tools agents use

• Using the console

  The inbox, memories and call history

• Managing connection tokens

  Expiry, reissue, revoke, and what to do after a leak

Last updated Oct 28, 2026