Use Flora from ChatGPT, Claude, and other LLM assistants
Connect your own AI assistant to Flora via MCP. Available only on paid plans.
Flora is great at generating flashcards from information on the public internet, or from documents that you manually upload. However, these days, a lot of useful knowledge is bound up in private conversations between large language models (LLMs) and their users, which Flora can’t easily access through the usual routes.
If you regularly work with an LLM assistant, you might want to connect it to your Flora library, allowing it to add documents, write questions, or have Flora write questions in bulk automatically.
Flora has an MCP server for exactly this reason, allowing you to access your account through assistant. The address is:
https://api.flora.day/mcp
What your assistant can do
An authorised LLM can read anything in your account which you grant access to - decks, sources, cards. They can add new sources via links, or as raw text, in a brand new deck or inserting them into an existing deck. They can write cards and add them to those decks manually, or hand over the responsibility for writing cards to Flora. It can also tidy up for you, looking through your decks, renaming them, editing or deleting cards, sources or decks according to your instructions.
The one thing your assistant can’t do is practise for you. That (obviously) needs to be done by you, in the Flora app!
Examples
- “Spin up a new Flora deck from this report.”
- “Add this article to my Flora deck on classification metrics, and get Flora to write a set of cards.”
- “Write 10 Flora cards in a new deck on what we just worked out”
- “Tidy up the cards in my Flora deck on Eric Hobsbawm - get rid of anything about his personal life, and concentrate on his published writing”
Connect ChatGPT
ChatGPT adds Flora in developer mode, on the web, with a Plus, Pro, Business, Enterprise or Edu plan. Turn on Developer mode in Settings, under Security and login. Then open Plugins, choose +, then Create MCP App. Call it Flora, paste the address above, choose OAuth, and sign in to Flora when ChatGPT opens it.
In a chat, choose Developer mode from the + menu and pick Flora. ChatGPT asks before Flora changes anything. It works in ChatGPT on the web, not yet in the phone app. On Business, an admin or owner turns developer mode on; on Enterprise and Edu, an admin has to allow it first.
Connect Claude
On claude.ai or in the Claude desktop app, open Customize, then Connectors, choose +, then Add custom connector. Call it Flora, paste the address above, and sign in to Flora when Claude asks. Or, signed in to Claude, open this link, which fills the form in. Once it’s added, Flora works in the Claude app on your phone too.
In a chat, turn Flora on from the + menu, under Connectors. On a Team or Enterprise plan, an owner adds Flora for everyone under Organization settings, then Connectors, and each person then chooses Connect.
In Claude Code, run this in a terminal:
claude mcp add --transport http --scope user flora https://api.flora.day/mcp
Then type /mcp in Claude Code and choose Flora to sign in. If you’ve already added Flora on claude.ai, Claude Code signed in to the same account has it too.
Connect another assistant
Any assistant that takes an MCP server’s address connects the same way: give it the address above, and sign in when it opens Flora. In Cursor, that’s an entry in mcp.json:
{
"mcpServers": {
"flora": { "url": "https://api.flora.day/mcp" }
}
}
What it costs
MCP access is a feature available to all paid accounts, as part of the regular monthly/annual subscription.
Disconnecting
Your Account page lists every assistant which is currently authorised to access your account. Disconnecting stops it at once, but any cards/decks you’ve added through that connection will stay until you manually delete them.
Privacy
Conversations between you and your LLM are none of our business! We only see the incoming requests to eg add a new deck, delete a card, etc, coming from your account, and Flora will only share your decks etc when asked.
See the privacy page for more detail about data what we store, and what we don’t.
Last updated 28 September 2026