Mind Archive

Own your AI memory.

Mind Archive keeps your AI conversations, notes and files as ordinary readable documents on your own computer — whichever AI you use next.

Get started Read the docs
0accounts to create
0bytes sent anywhere
1folder you can open
.mdfiles you can read

Import it, search it, keep it.

Three things, done properly. Everything else is deliberately absent.

Import

Drop an export into the inbox folder and it imports itself. ChatGPT and Claude are supported today, as .zip or raw .json. Each provider sits behind its own adapter, so adding one never touches the rest.

Search

Full-text search across everything you have imported, with the matching line shown in place. Tags narrow it further. The index is SQLite and can be rebuilt from the files at any time.

Keep

Your archive is Markdown and JSON in a normal folder. Delete Mind Archive tomorrow and every conversation still opens in any editor. Export the whole thing as a zip whenever you like.

Get started

Docker is the supported path, because it avoids every "which Python do I have" problem. Native setup is in the development guide. 

git clone https://github.com/narainsagar/mindarchive.git
cd mindarchive
cp .env.example .env

python scripts/dev.py up --build
  1. Export from your AI. In ChatGPT: Settings → Data controls → Export data. ChatGPT emails you when it is ready, and says this can take a few days — the download link then expires 24 hours after that email arrives.
  2. Drop it in the inbox. Save the file into data/inbox/ and Mind Archive picks it up, or choose it in the interface.
  3. Read it as documents. Each conversation becomes one Markdown file, dated and titled. 
data/
├── archive/
│   ├── 2026/
│   │   ├── chatgpt/
│   │   │   ├── 2026-03-11-postgres-index-strategy.md
│   │   │   └── 2026-03-14-reading-list.md
│   │   └── claude/
│   │       └── 2026-04-02-mindarchive-naming.md
│   └── index.sqlite            metadata only — rebuildable
└── inbox/                      drop exports here

The interface runs at http://localhost:5173 and the API at http://localhost:8000. Both stay on your machine.

What "local-first" actually means here

Not a marketing word. There is no account, no sync and no telemetry. The API binds to localhost and has no authentication, because it is a single-user application that never leaves your computer. Cloud backup is optional, off by default, and cannot be enabled silently.

Your archive lives in a folder you chose, as a bind mount rather than a Docker volume — so no cleanup command in this project can delete it. Read the security notes before you trust it with anything, and read the source while you are there.

Proprietary — all rights reserved

Mind Archive is proprietary software. It is not open source and it is not source-available, and describing it as either would be wrong. No licence to use, copy, modify or distribute it is granted.

Nothing is for sale, and there is no price list. Any future licensing model will be published here when it exists, not before. Read the licensing page.

What this does not change: your archive is ordinary Markdown and JSON on your own disk, exportable at any time. Licensing governs the software, never your data.

Documentation