Rember MCP

🚀 Rember MCP

Allow Claude to create flashcards on Rember with the official Model Context Protocol (MCP). Rember helps you learn and remember anything you care about by scheduling spaced repetition reviews.

✨ Features and Examples:

• Create flashcards from chat history "...I like your answer, help me remember it"
• Create flashcards from PDFs "Create flashcards from Chapter 2 of this PDF"

📦 Installation

To run the Rember MCP server using npx, execute the following command:

npx -y @getrember/mcp --api-key=YOUR_REMBER_API_KEY

Replace YOUR_REMBER_API_KEY with your actual Rember API key, which can be found on your Settings page. The API key should follow the format rember_ followed by 32 random characters.

Usage with Claude Desktop

Add the following content to your claude_desktop_config.json. For more details, refer to here.

{
  "mcpServers": {
    "rember": {
      "command": "npx",
      "args": ["-y", "@getrember/mcp", "--api-key=YOUR_REMBER_API_KEY"]
    }
  }
}

💻 Available Tools

• create_flashcards: Create flashcards using AI. This tool creates flashcards from Claude's note list and then calls the Rember API to generate several flashcards for each note. After learning something new in a chat with Claude, you can ask "Help me remember this", "Create a few flashcards", or "Add to Rember".

🔧 Technical Details

Here are some best practices we've summarized during the development of the Rember MCP server:

• Log to stderr early, which is crucial for debugging.

• First, create a simple MCP tool and verify that Claude can call it correctly.

• Invest time in iteratively improving the tool description:

  • Include detailed information about your product and its URL. This serves two purposes: helping Claude use the tool correctly and allowing Claude to answer user questions about the product.
  • Clearly explain what MCP is. In some cases, Claude may mistakenly think MCP stands for "Multiple Choice Prompt", which is concerning.
  • Describe the tool input in detail.
  • Explain what happens after calling the tool. We explicitly state that the input note array will be sent to the Rember API to generate flashcards for each note.
  • Provide specific examples of using the tool (e.g., "Create flashcards from a conversation with Claude", "Create flashcards from a PDF") and give Claude specific usage instructions.
  • Include a list of rules to guide Claude in using the tool appropriately.

• Strategically use tool call responses, which are not directly shown to the user but interpreted by Claude:

  • In case of success, the Rember API does not return the number of created flashcards. Claude only knows the number of Rember items created. We clarify this because otherwise, Claude tends to incorrectly guess the number of created flashcards.
  • For users who have reached their monthly limit, we instruct Claude to inform them about Rember Pro subscription options and the relevant URL.

• Implement a retry mechanism for transient errors and set appropriate timeouts.

• We've collected enough edge cases to make manual testing on Claude Desktop cumbersome. We created a test suite to perform manual checks and include some simple assertions by simulating Claude Desktop behavior and calling the Claude API (with system prompts from claude.ai).

Missing Features

• Monitoring and observability. Currently, we have no idea about the occurrence of errors.
• More comprehensive error handling.
• More iterative improvements to tool descriptions.
• More automated tests.