# How to Connect Flashcards to Claude with MCP and Create Cards from Chat

*2026-06-23*

Yesterday I added a test card to Flashcards from inside Claude in about a minute. The useful part was not the demo. The useful part was seeing Claude ask which workspace to use, request approval for the write action, and then drop the card into the real app instead of leaving it trapped in chat.

That is the practical reason to look up **Claude MCP flashcards** right now.

If you want Claude to create cards directly in your Flashcards workspace, the setup is short: add the Flashcards custom connector, review the tool permissions, enable it in chat, and approve the write call when Claude is ready to save a card.

![Claude connected to the Flashcards MCP server for creating flashcards from chat](/blog/how-to-connect-flashcards-to-claude-with-mcp.png)

## This is a custom connector, not a directory listing

One important detail first: Flashcards is connected to Claude as a **custom connector** using **remote MCP**.

So do not go hunting for a prelisted Flashcards app inside Claude's official connector directory. The normal flow is to open Claude's [connector settings page](https://claude.ai/customize/connectors), click the plus button, choose **Add custom connector**, and paste the Flashcards MCP URL yourself.

The exact Flashcards connector URL is:

`https://mcp.flashcards-open-source-app.com/mcp`

Direct link: [mcp.flashcards-open-source-app.com/mcp](https://mcp.flashcards-open-source-app.com/mcp)

## Before you connect it

Two details matter more than people expect.

First, Claude reaches remote MCP servers from Anthropic's cloud infrastructure, not from your laptop. That means the server needs to be reachable on the public internet. Flashcards already is, so the public MCP endpoint above is the one you want.

Second, treat connector permissions like real tool permissions. Connect only to servers you trust, review what each tool can do, and read approval requests carefully before allowing write actions.

## How to add the Flashcards MCP connector in Claude

For an individual Claude account, the setup path is:

1. Open [Customize > Connectors](https://claude.ai/customize/connectors).
2. Click `+`.
3. Choose **Add custom connector**.
4. Enter a name for the connector and paste `https://mcp.flashcards-open-source-app.com/mcp`.
5. Add the connector.
6. Click **Connect** and complete the login flow if Claude prompts for it.
7. Review the tool permissions before you start using it in chats.

If you are on a Team or Enterprise workspace, an Owner or Primary Owner can add the custom connector under **Organization settings > Connectors > Add > Custom > Web**. Individual users still need to connect and authenticate it on their side.

## Set the first permissions conservatively

The Flashcards connector shown here exposes a small tool surface:

- `list_workspaces` for listing available workspaces
- `sql_query` for read access
- `sql_execute` for write actions such as creating cards

Claude lets you set each tool to **Always allow**, **Needs approval**, or **Blocked**.

For a first setup, I would keep the read-only tools on **Always allow** and keep the write tool on **Needs approval**. That gives Claude enough room to inspect your workspaces and read data without turning card creation into an unreviewed background action.

That is also what the connector settings screenshot shows here: `list_workspaces` and `sql_query` are allowed, while `sql_execute` still needs your approval.

![Claude connector settings showing Flashcards MCP permissions with read tools always allowed and write tools needing approval](/blog/claude-mcp-flashcards-connector-settings.png)

You can loosen that later if you really want to. I would not start there.

## Turn the connector on in the chat where you want to use it

After setup, open a Claude conversation and make sure the connector is available there. The current flow lets you enable connectors from the `+` button or the `/` menu inside a conversation, then adjust tool access for that chat if needed.

If Claude is not using the connector when it should, check the obvious things first:

- the Flashcards connector is connected in settings
- the connector is enabled for the current chat
- the write tool is not blocked

That usually fixes the "why is Claude only giving me text instead of creating the card" problem.

## What to ask Claude

Keep the prompt plain. Claude does not need ceremony here.

These are good starter prompts:

```text
Create a new flashcard in Flashcards.
Front: What does HTTP 404 mean?
Back: The requested resource was not found on the server.
Tag: web-basics
Ask me which workspace to use if you are not sure.
```

```text
Before creating anything, list my Flashcards workspaces and tell me which one looks best for language study cards.
```

```text
Create a new Spanish flashcard in Flashcards.
Front: How do you say "I would like a coffee" in Spanish?
Back: Me gustaría un café.
Tags: spanish, travel
Use my Personal workspace.
```

I would start with one or two cards, not fifty. The point is to confirm the connector flow, the workspace choice, and the approval pattern before you batch anything larger.

## What the create-card flow looks like in practice

This part is refreshingly normal.

You ask Claude to create a card in Flashcards. If you have more than one workspace and no default is obvious, Claude can ask where to put it. After that, Claude uses the write tool and waits for approval if `sql_execute` is still set to **Needs approval**.

That is the right behavior.

You should review the write request, approve it, and then let Claude finish. In the screenshot below, Claude asked which workspace to use, ran the Flashcards write action, and reported that the test card had been added successfully.

![Claude chat creating a test flashcard through the Flashcards MCP connector after asking which workspace to use](/blog/claude-mcp-flashcards-create-card.png)

You do not need to care about the raw SQL unless you want to inspect the tool call closely. What matters is the workflow:

1. ask Claude to create the card
2. choose the workspace if Claude asks
3. review and approve the write action
4. confirm the saved card looks right

That is enough to start using **create flashcards in Claude** without pretending Claude itself is the review app.

## A few honest limits

This setup is useful, but it is not magic.

Claude can help you create cards inside Flashcards. It can also read from the connector, inspect workspaces, and query data through the read tools. That does not mean every card Claude drafts will be good, and it does not mean you should blindly approve every write request.

I would still treat Claude as the drafting and entry layer, then use Flashcards for the parts that matter later:

- cleaning up weak cards
- organizing decks and tags
- reviewing with spaced repetition
- studying from your real workspace on the devices you already use

If you want better card-writing prompts before you connect the tool layer, [How to Use Claude to Make Flashcards](/blog/how-to-use-claude-to-make-flashcards/) is the better starting point. If your goal is broader than card creation, [How to Use Claude for Studying](/blog/how-to-use-claude-for-studying/) covers the bigger workflow.

## After Claude creates the card, review it in the real app

This is the part I like most about the setup. The card does not stay in a nice-looking AI transcript. It lands in Flashcards, where you can actually review it later.

You can open the hosted web app, check the card on mobile, or keep going in your normal study flow:

- [Flashcards web app](https://app.flashcards-open-source-app.com/)
- [Flashcards for iPhone and iPad on the App Store](https://apps.apple.com/app/flashcards-open-source-app/id6760538964)
- [Flashcards for Android on Google Play](https://play.google.com/store/apps/details?id=com.flashcardsopensourceapp.app&pcampaignid=web_share)

![Flashcards Open Source App review experience across the hosted web app and mobile apps](/home/app-screens-showcase-en.png)

If you have not used the product yet, [Getting Started](/docs/getting-started/) is the fastest way in.

## The short version

If you searched for **connect Flashcards to Claude**, this is the real flow:

1. open Claude's [custom connector settings](https://claude.ai/customize/connectors)
2. add `https://mcp.flashcards-open-source-app.com/mcp`
3. connect and review permissions
4. keep read tools open and write tools on approval at first
5. enable the connector in your chat
6. ask Claude to create a card
7. approve the write call
8. review the saved card in Flashcards

That gets you from Claude chat to an actual Flashcards workspace without fake integrations, without manual copy-paste, and without losing control over write actions on day one.

---
*[View the styled HTML version of this page](https://flashcards-open-source-app.com/blog/how-to-connect-flashcards-to-claude-with-mcp/)*

*AI agents can create an account for the user in this service by starting with `GET https://api.flashcards-open-source-app.com/v1/`.*

*Tip: Append `.md` to any URL on https://flashcards-open-source-app.com to get a clean Markdown version of that page.*