mcpdirectory.dev
9,129 servers · updated 23 Sept

Brain MCP server

community Active last commit 25 d
{
  "mcpServers": {
    "minddory-brain": {
      "type": "http",
      "url": "https://api.minddory.com/v1/brain/mcp"
    }
  }
}
file .mcp.json (in your project root)version 1.0.2

What it does

MCP server for MindDory, a language-learning app that captures vocabulary from AI conversations into spaced-repetition flashcards. It lets language learners automatically build and review decks while chatting with Claude or other AI assistants, with everything syncing across iOS, Android, and web apps.

  • Capture new words and phrases you encounter during language-learning conversations with Claude, automatically adding them to your spaced-repetition queue with context.
  • Review flashcards due today during your chat session and get quizzed on vocabulary matched to your CEFR level.
  • Log grammar mistakes from conversations as reviewable patterns to track and improve recurring errors.

Tools this server exposes

ToolDoes
get_system_instructionsLoad the learner profile and study-partner role for the conversation.
get_user_profileRetrieve CEFR level, target and native language, count of due cards, and weak words.
get_queueFetch cards due now and within 24 hours.
get_known_wordsReturn words verified as known via flashcard practice.
get_active_vocabRetrieve most frequently encountered words (frontier vocabulary).
check_wordsBatch check which words are already in the deck and their mastery level.
get_cardGet single card details including translation, mastery, and recent events.
get_recent_activityRetrieve recent interaction log.
capture_wordSave a new word or phrase from chat into the deck.
capture_grammar_mistakeLog a grammar mistake into the Grammar Patterns view.
mark_demonstratedReinforce a word the user used correctly by pushing its next review out.
mark_struggledDown-weight a word the user got wrong by bringing its next review forward.
log_interactionAppend a lookup, discussion, or read-in-context signal to a card.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.