Overview

Rember MCP is a lightweight MCP (Model Context Protocol) server that connects Claude (or any MCP-compatible model client) to the Rember spaced-repetition service. It exposes a small set of tools that let a model create and enqueue flashcards (rembs) in your Rember account so chat sessions, documents, or other context can be converted into scheduled review items automatically.

This server is intended for developers who want to add recall and spaced-repetition workflows to conversational agents. By running the MCP server locally or as a subprocess, Claude can call a tool such as create_flashcards with a list of notes; the server then forwards those notes to the Rember API and reports back success/failure. The MCP implementation follows the Model Context Protocol so it integrates with MCP-compatible clients such as Claude Desktop.

Features

Create flashcards (rembs) from chat content, document extracts, or other extracted notes
Integrates with the official Rember API for scheduled spaced-repetition reviews
Simple MCP tool surface for easy integration with Claude Desktop and other MCP clients
Example configuration for running via npx (no build step required)
Guidance and defensive defaults for retries, logging, and helpful tool descriptions

Installation / Configuration

Run the MCP server directly with npx. Replace YOUR_REMBER_API_KEY with an API key from your Rember account (found in Settings → MCP API). Keys use the format rember_<32 chars>.

# Run the Rember MCP server with your API key
npx -y @getrember/mcp --api-key=YOUR_REMBER_API_KEY

To register the MCP server with Claude Desktop, add an entry to your claude_desktop_config.json:

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

Details on the Model Context Protocol and client configuration are available at https://modelcontextprotocol.com/. Rember account settings (to retrieve the MCP API key) are at https://rember.com/settings/mcp-api.

Available Tools

create_flashcards

Purpose: Accepts an array of notes and asks Rember to generate one or more flashcards for each note.
Input: JSON object with a notes array. Each note is typically a short text snippet or summary extracted by the model.
Output: A tool response indicating success or failure. On success the Rember API returns rembs added to the account; the MCP server returns a structured response that Claude can interpret (the API does not guarantee a flashcard count, so the server intentionally reports remb creation rather than exact card counts).

Example input (conceptual):

{
  "notes": [
    "Photosynthesis converts light energy into chemical energy in chloroplasts.",
    "Newton's second law: F = ma, acceleration proportional to net force."
  ],
  "source": "chat:session-123"
}

Example conceptual response:

{
  "status": "ok",
  "message": "Notes sent to Rember. Created rembs for N notes (Rember returns remb IDs)."
}

Use Cases

From a Claude chat:
- Prompt: “I like your answer — help me remember it.” Claude can summarize the important points, call the create_flashcards tool with those summaries, and tell the user that the items were added to Rember for spaced-repetition review.
From a PDF or Document:
- Workflow: Extract chapter text (server or client-side), ask Claude to summarize or distill key facts per paragraph, then call create_flashcards with the resulting note array: “Create flashcards from chapter 2 of this PDF.”
Classroom or course material:
- Feed lecture transcripts or slide bullets as notes. Claude can convert each bullet into a question/answer pair and send them to Rember.
Incremental review:
- During a long tutoring session, periodically prompt Claude to “add these 5 takeaways to Rember” so they appear in the user’s review schedule.

Concrete prompt example for Claude:

“Summarize the last message into three concise facts and add them to Rember for review.”

Development notes & best practices

Log to stderr early for easier debugging when MCP clients spawn the server.
Start with a minimal MCP tool and verify tool calls before adding complexity.
Write a rich tool description: explain MCP, your product URL, input schema, allowed use cases, and post-call behavior (what the user will see).
Handle transient errors with retries and sensible timeouts; inform the model about rate limits or monthly limits so it doesn’t hallucinate success counts.
Tests: simulate client behavior (e.g., Claude Desktop prompts) to validate tool prompts and edge cases.
Known gaps: limited telemetry/observability, more exhaustive error handling and more unit tests.

Repository and source code: https://github.com/rember/rember-mcp Rember: https://rember.com/ | MCP spec: https://modelcontextprotocol.com/

Rember MCP Server for Spaced Repetition