How to Transfer Your ChatGPT Memory to OpenClaw

How OpenClaw memory actually works

Before you import anything, it helps to know what you're dealing with. OpenClaw doesn't have a cloud memory system like ChatGPT does. Instead, it stores everything as plain text files on your own computer, specifically in a folder called your workspace.

The main file you care about is MEMORY.md. Whatever you put in there gets loaded into every conversation you have with your agent. It's simple, transparent, and you can edit it with any text editor.

The downside is that recall isn't automatic. Unlike ChatGPT's memory system, which pretty reliably surfaces what it knows about you, OpenClaw is more of a best-effort situation. The model looks at your memory files when it thinks they're relevant, which is most of the time, but not always. Think of it as a very good assistant who occasionally forgets to check their notes.

Why you need to trim before you import

Your first instinct might be to paste your entire ChatGPT export into MEMORY.md and call it done. That technically works, but it gets expensive fast. OpenClaw reads that file on every single message you send. Every word in there costs you tokens, even if 90% of it isn't relevant to what you're asking right now.

People who've tried the "dump everything" approach usually come to the same conclusion: treat MEMORY.md like a business card, not a biography. Name, location, a few key projects, core preferences. The stuff that's almost always relevant no matter what you're talking about. Everything else is just noise that slows down responses and burns through your API budget.

The sweet spot is around 20-30 entries. That's enough for your agent to know who you are without costing you extra on every conversation.

Step 1: Export your ChatGPT memory

Open any ChatGPT conversation and paste in the prompt below. It asks ChatGPT to dump everything it has stored about you in a clean, copy-friendly format.

You can also double-check what's stored by going to Settings > Personalization > Manage Memories. ChatGPT has two layers of memory: things you explicitly asked it to save, and preferences it picked up from your conversation history. The prompt below tries to capture both.

If ChatGPT says it has no memories, check that the Memory toggle is turned on under Settings > Personalization.

Step 2: Trim it in the ImportMemory editor

ChatGPT might hand you back 50, 80, even 100 entries. You don't want all of that in MEMORY.md. Paste the raw export into the ImportMemory editor and go through it.

Keep anything that's still true and likely to come up regularly. Cut things like:

• Old job titles or companies you've left
• Completed projects you're no longer working on
• One-off requests that aren't part of your regular workflow
• Duplicate or near-duplicate entries

Aim for a tight list of 20-30 facts. When you're done, copy the output.

Step 3: Paste it into MEMORY.md

Open your OpenClaw workspace folder. By default it lives at ~/.openclaw/workspace/. Open MEMORY.md in any text editor and paste your trimmed memory in:

## Identity
- Name: [your name]
- Location: [city, country]
- Job: [role at company]

## Active Projects
- [Project name]: [one line description]

## Preferences
- Always respond in bullet points for technical questions
- Keep answers concise unless I ask for detail
- I use Python and TypeScript primarily

Save the file and restart your OpenClaw gateway with openclaw gateway. Your agent will pick it up on the next conversation.

Tips for a better result

• Keep MEMORY.md lean. 20-30 entries is the sweet spot. More than that and you're wasting tokens on every message.
• Use clear section headers. Grouping by Identity, Projects, and Preferences makes it easier for the model to parse.
• Review it every few weeks. It's a plain text file. Update it any time.
• Don't paste sensitive info. Your memory content still gets sent to whatever LLM API you're using.
• Recall isn't guaranteed. If something important keeps getting missed, try stating it directly in conversation.

Frequently asked questions