scrimba-teaching-mcp 3.0.5

Unified MCP server for interactive programming education with Scrimba's revolutionary methodology

Scrimba Teaching MCP System

mcp-name: io.github.srikrishna235/scrimba-teaching-mcp

A revolutionary teaching system implementing Per Borgen's Scrimba methodology, available as MCP servers for both Claude Desktop and Claude CLI.

🚀 v1.1.0 - Agent-Router Update!

NEW! scrimba_agent - Unified tool that embeds agent personalities and auto-routes requests!

• Analyzes user intent automatically
• Routes to visual, interactive, project, or challenge modes
• No need for separate agents - everything in one tool!

Architecture

┌─────────────────────┐     ┌────────────────────┐
│  Teaching MCP       │     │  CLI Wrapper MCP   │
│  (Pure Logic)       │     │  (Agent Support)   │
├─────────────────────┤     ├────────────────────┤
│ • scrimba_agent()   │     │ • execute_with_    │
│   (NEW! v1.1.0)     │     │   agent()          │
│ • teach()           │     │ • list_agents()    │
│ • give_challenge()  │     │                    │
│ • check_code()      │     │                    │
│ • visualize()       │     │                    │
│ • start_project()   │     │                    │
└─────────────────────┘     └────────────────────┘
         ↑                           ↑
         │                           │
    Claude Desktop              Claude CLI
    (Direct MCP)              (With Agent Support)

Features

Core Scrimba Methodology

• 60-Second Rule: Users coding within 60 seconds
• Micro-Lessons: [HOOK: 20s] → [CONCEPT: 60s] → [CHALLENGE: 120s] → [CELEBRATE: 10s]
• 5-Level Progression: Every concept taught in 5 progressive steps
• Console.log Driven Development: Verify everything immediately
• Error-First Learning: Celebrate mistakes as learning moments
• Real Projects: Passenger Counter, Blackjack Game, Chrome Extension

Teaching Tools

• teach(topic, step) - Interactive lessons with Scrimba methodology
• give_challenge(difficulty) - Immediate coding challenges
• check_code(code) - Validation with encouragement
• next_lesson() - Progress through 5-level curriculum
• start_project(name) - Build real applications
• visualize_concept(concept) - Generate visual learning materials
• show_progress() - Track learning journey

Installation

For Claude Desktop

1. Install the MCP server:

cd scrimba-mcp-unified
pip install -e .

2. Configure Claude Desktop: Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "scrimba-teaching": {
      "command": "python",
      "args": ["/path/to/teaching-server/teaching_mcp.py"]
    }
  }
}

3. Start using in Claude Desktop:

teach me variables
give me a challenge
check my code: let age = 25

For Claude CLI

1. Install both servers:

cd scrimba-mcp-unified
pip install -e .

2. Register MCP servers: The .mcp.json file is already configured. Claude CLI will automatically detect both servers.

3. Option A - Direct MCP Usage:

# Use MCP tools directly
claude "teach me variables"
claude "give me a coding challenge"

4. Option B - With Agent Orchestration:

# Create teaching agents
claude "create teaching orchestrator agent"

# Use agents for orchestration
claude @teaching-orchestrator "teach me variables"

Usage Examples

Direct Teaching (Claude Desktop & CLI)

User: teach me variables
Bot: [Full Scrimba micro-lesson with story, concept, and challenge]

User: let myAge = 25
Bot: [Celebration and next challenge]

User: next lesson
Bot: [Level 2 of 5-level progression]

Project-Based Learning

User: start project passenger_counter
Bot: [Per's subway story + starter code + 5 build steps]

User: show my progress
Bot: [Visual progress bar + achievements + motivation]

Visual Learning

User: visualize variables
Bot: [Detailed image prompt for educational visualization]

Key Differences

Feature	Claude Desktop	Claude CLI
Direct MCP Tools	✅ Yes	✅ Yes
Agent Support	❌ No	✅ Yes (via wrapper)
Installation	Single server	Both servers (optional)
Usage	Direct tools only	Tools or agents

Development

Testing the Teaching Server

cd teaching-server
python teaching_mcp.py --test

Testing with Claude CLI

# Test direct MCP
claude "teach me functions" --mcp-debug

# Test with agents
claude @teaching-orchestrator "teach me functions"

Scrimba Methodology Details

Micro-Lesson Structure

Every lesson follows this exact timing:

• 0-20s: Personal story hook ("When I was 19...")
• 20-80s: ONE concept only
• 80-200s: Immediate practice
• 200-210s: Celebration

Progressive Complexity (5 Levels)

1. Basic declaration
2. Modification/reassignment
3. Shorthand syntax
4. Advanced usage
5. BUILD REAL APP!

Language Patterns

• ALWAYS: "Hey buddy!", "This is HUGE!", "Super common!"
• NEVER: "Let me explain...", "This is wrong", "As you should know..."

Projects

1. Passenger Counter: Variables & Functions
2. Blackjack Game: Logic & Conditionals
3. Chrome Extension: Advanced concepts

Contributing

The teaching logic is centralized in teaching_mcp.py. To add new lessons:

1. Add to LESSONS dictionary with 5 levels
2. Follow exact Scrimba structure
3. Include personal stories
4. Test with both Desktop and CLI

License

MIT - Share the Scrimba way of learning!

Credits

Based on Per Borgen's revolutionary teaching methodology from Scrimba. "The only way to learn to code is to write a lot of code!"