LinguaMCP
Serves daily language lessons from a structured curriculum, enabling natural practice through conversation with AI tutors that track progress.
README
LinguaMCP
Open curriculum protocol for AI language tutors — learn naturally through daily practice with any AI tool.
What is it?
LinguaMCP is an MCP (Model Context Protocol) server that serves daily language lessons from a structured curriculum. It works with Claude, Cursor, Windsurf, or any AI tool that supports MCP.
Instead of flashcards or grammar drills, you practice naturally through conversation. Your AI tutor knows what you've learned, tracks your progress, and introduces new material when you're ready.
Quick Start
Option 1: Zero-config SQLite (recommended)
# Install globally
npm install -g @lingua/mcp
# Initialize database with curriculum
lingua-mcp-init
# Start the server
lingua-mcp
That's it. No database to set up, no API keys, no cloud services. Your progress lives in ~/.lingua-mcp/lingua.db.
Option 2: Supabase (cloud or self-hosted)
# Clone and install
git clone https://github.com/Marsmanleo/LinguaMCP.git
cd LinguaMCP
npm install
# Set up environment
cp .env.example .env
# Edit .env with your Supabase credentials
# Run the Supabase migration (in supabase/migrations/)
# Then ingest the curriculum
npm run ingest
# Start the server
npm start
Configure Your AI Tool
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"lingua": {
"command": "npx",
"args": ["-y", "@lingua/mcp"]
}
}
}
Cursor / Windsurf
Add to your MCP settings:
{
"lingua": {
"command": "node",
"args": ["/path/to/LinguaMCP/server.mjs"]
}
}
Supabase mode
Set environment variables before starting:
SUPABASE_BASE_URL=https://your-project.supabase.co \
SUPABASE_ANON_KEY=your-anon-key \
node server.mjs
MCP Tools
Once connected, your AI tool gets three new tools:
| Tool | Description |
|---|---|
get_today_lesson |
Returns your next unmastered lesson with context and practice prompts |
get_user_progress |
Shows mastery, practiced, and remaining counts |
log_response |
Records your practice response (score 1-5, ≥4 = mastered) |
Curriculum
Currently using structured content from English-level-up-tips (50k+ stars, MIT License).
Available skill books:
- 🔗 Linking Words — 4 chapters, 20 lessons covering contrast, addition, cause-effect, and condition connectors
More skill books and languages coming soon.
How It Works
┌─────────────┐ MCP ┌─────────────┐ Query ┌──────────┐
│ Your AI │──────────────▶│ LinguaMCP │──────────────▶│ Database │
│ Tool │◀──────────────│ Server │◀──────────────│ │
└─────────────┘ Lessons & └─────────────┘ Progress & └──────────┘
Progress Lessons
- You chat with your AI tool normally
- The AI calls LinguaMCP to get your lesson for the day
- You practice through conversation
- LinguaMCP tracks what you've mastered and what's next
Architecture
server.mjs ← MCP server entry point
storage/
interface.mjs ← Storage contract
supabase.mjs ← Supabase/PostgREST adapter
sqlite.mjs ← SQLite adapter (zero-config)
schema/
sqlite.sql ← SQLite schema
scripts/
ingest-english-tips.ts ← Content ingestion (Supabase)
init-sqlite.mjs ← Content ingestion (SQLite)
src/
index.ts ← TypeScript types
Storage Modes
| Mode | When | Database | Best for |
|---|---|---|---|
| SQLite | No SUPABASE_BASE_URL set |
~/.lingua-mcp/lingua.db |
Personal use, self-hosted |
| Supabase | SUPABASE_BASE_URL is set |
PostgreSQL via PostgREST | Cloud, multi-user, teams |
Same tools, same responses, same curriculum. The server auto-detects which mode to use.
Contributing
See CONTRIBUTING.md for how to add curriculum content, storage adapters, or language support.
License
Apache License 2.0 — see LICENSE.
Curriculum content from English-level-up-tips is MIT licensed.
Related
- MarsNMe — AI memory companion (sister project)
- Model Context Protocol — The protocol that makes this work
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。