Chatgpt Migrator

Switch from ChatGPT to Claude

Migrate your entire ChatGPT conversation history to Claude. The skill analyses everything you've discussed with ChatGPT and creates:

• A personal profile so Claude knows who you are from day one
• Memories pulled from your conversations (30-100+ facts about you)
• Custom instructions tailored to how you like to work
• A topic index of everything you've talked about, organised by theme
• Skill and integration suggestions based on what you actually use AI for

Everything is ready to import into Claude when it's done.

---

What you get

| Output | What's in it | |---|---| | Personal profile | A comprehensive "about you" document covering who you are, what you do, your interests, communication style, and key life context. Add it to a Claude Project or keep it in a folder Cowork can access, and Claude has your full context from the first message. | | Memories | 30-100+ discrete facts extracted from your conversations, ready to add to Claude's memory. Job title, location, preferences, current projects, personal details - each one a single, specific, useful fact. | | Custom instructions | A tailored system prompt you can paste straight into Claude's settings. Encodes your communication preferences, key context, and specific instructions so Claude works the way you like from day one. | | Skill suggestions | 3-5 custom Cowork skills based on how you actually used ChatGPT. Each includes a name, what it would do, why it suits you, and enough detail to build it. Weekly reporting, data analysis, content drafting - your patterns become purpose-built skills. | | Integrations & MCP servers | Personalised recommendations for official MCP servers and Claude connectors based on the tools in your ChatGPT history. Google Workspace, GitHub, Slack, Notion, Figma, Todoist, databases, and more - but only the ones that match your actual usage. Each recommendation explains what it is, why it's relevant to you, and what it enables. | | Topic index | Your entire conversation archive organised by theme. For each topic: a brief description, how many conversations touched on it, the date range, and the most notable conversation titles. | | Migration summary | Overview of what was processed, what was found, and suggestions for next steps. The first thing you'll read when the migration finishes. |

---

Get started

Step 1: Request your ChatGPT export

1. Go to chatgpt.com and click your profile icon (top-right corner)
2. Click Settings
3. Click Data controls, then Export data, then Confirm export
4. OpenAI will email you a download link (usually within an hour, sometimes up to 24 hours)
5. When the email arrives, download the zip file and unzip it somewhere you can find it

> [!TIP] > While you wait for the export, you can do a quick partial migration right now. See [Don't have your export yet?](#dont-have-your-export-yet) below.

Step 2: Install the skill

You need a Claude Pro, Team, or Enterprise plan.

1. Click here to download the skill file - your browser will save a file called chatgpt-migrator.skill
2. Open the Claude desktop app
3. Go to Settings > Customize > Skills
4. Click Upload skill
5. Drag the chatgpt-migrator.skill file from your Downloads folder into the upload area
6. You'll see it appear in your skills list - that's it, it's installed

Step 3: Run the migration

1. In Cowork, start a new session
2. Add your unzipped ChatGPT export folder to the session (drag the folder in, or use the attachment button)
3. Type "migrate my ChatGPT data" and press enter
4. The skill walks you through everything from here, with progress updates along the way

It will ask you a few questions about what you want to focus on, then process your conversations. When it's done, it helps you import everything into Claude.

---

Don't have your export yet?

You can do a quick partial migration while you wait:

1. Open [STANDALONE-PROMPT.md](STANDALONE-PROMPT.md) and copy the entire prompt
2. Paste it into a new ChatGPT conversation and send it
3. ChatGPT will output a block of memories - copy the whole response
4. Go to claude.com/import-memory and paste it in

This grabs what ChatGPT has memorised about you so far. When your full export arrives, the skill will cross-reference and build on these.

---

How long does it take?

| Export size | Estimated time | |---|---| | Under 200 conversations | 10-15 minutes | | 200-800 conversations | 20-30 minutes | | 800-1500 conversations | 30-45 minutes | | 1500+ conversations | 45-60 minutes |

You get progress updates throughout. If a session runs out of context, start a new one and say "continue" - it picks up where it left off.

Privacy

Your files stay on your machine. Conversation content is sent to Claude for processing, subject to Anthropic's privacy policy. Nothing is sent to any third-party services.

---

Troubleshooting

| Problem | Solution | |---|---| | The skill file opened in my browser instead of downloading | Right-click the download link and choose Save link as... or Download linked file | | Cowork says the skill file is invalid or can't read it | Make sure you're uploading the .skill file, not the README or a zip of this page. Re-download it from the [link in Step 2](#step-2-install-the-skill) | | The session ran out before finishing | Start a new Cowork session, add the same export folder, and say "continue" | | Can't find the skill in Cowork's search | This skill isn't in the public registry. Install it by uploading the file as described in [Step 2](#step-2-install-the-skill) |

---

Using Claude Code instead of Cowork?

git clone https://github.com/JamsusMaximus/chatgpt-migrator.git
cd chatgpt-migrator

Then open your ChatGPT export folder and say "migrate my ChatGPT data". The skill is automatically available.

macOS note: If you see a popup asking to install Command Line Tools, click Install, wait a few minutes, then run the commands above again.

Developer information

Requirements

• A ChatGPT data export ([how to get one](#step-1-request-your-chatgpt-export))
• Python 3.8+ (for the preprocessor script - install via python.org if you don't have it)
• Claude Cowork or Claude Code

Repo structure

chatgpt-migrator.skill      # Upload this to Cowork (self-contained package)
SKILL.md                    # Skill instructions and YAML frontmatter
scripts/
  preprocessor.py           # Compresses the raw export into batches
references/
  batch-analysis-prompt.md
  synthesis-prompt.md
evals/
  evals.json
STANDALONE-PROMPT.md        # Copy-paste prompt for quick partial migration
LICENSE                     # Apache 2.0
README.md                   # This file

How the pipeline works

1. Identify the export and check for any previous progress
2. Interview you about scope, priorities, and target platform
3. Preprocess the raw export into batches (a Python script handles this in under a minute)
4. Analyse each batch in parallel using subagents (waves of 3-5 at a time)
5. Synthesise all analyses into the final output files
6. Fact-check up to 10 key facts with you, then walk you through importing into Claude

Progress is saved to progress.json after every step.

Licence

Apache 2.0. See [LICENSE](LICENSE).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: JamsusMaximus
• Source: JamsusMaximus/chatgpt-migrator
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.