mcmodding-mcp

MCP server providing AI assistants with comprehensive, up-to-date Minecraft modding documentation for Fabric and NeoForge.

What is this?

mcmodding-mcp is a Model Context Protocol (MCP) server that gives AI assistants like Claude direct access to Minecraft modding documentation. Instead of relying on potentially outdated training data, your AI assistant can search real documentation, find code examples, and explain concepts accurately.

Key Benefits

• Always Current - Documentation is indexed weekly from official sources
• Accurate Answers - AI responses backed by real documentation, not hallucinations
• Code Examples - Searchable code blocks with proper context
• Semantic Search - Understands what you mean, not just keywords
• Zero Config - Works immediately after installation

---

Quick Start

Installation

# Install globally
npm install -g mcmodding-mcp

Configure Your AI Client

Add to your MCP client configuration (e.g., Claude Desktop):

{
  "mcpServers": {
    "mcmodding": {
      "command": "mcmodding-mcp"
    }
  }
}

That's it! Your AI assistant now has access to Minecraft modding documentation.

---

Available Tools

The MCP server provides four powerful tools:

search_fabric_docs

Search documentation with smart filtering.

// Example: Find information about item registration
{
  query: "how to register custom items",
  category: "items",           // Optional filter
  loader: "fabric",            // fabric | neoforge
  minecraft_version: "1.21.4"  // Optional version filter
}

get_example

Get working code examples for any topic.

// Example: Get block registration code
{
  topic: "custom block with block entity",
  language: "java",
  loader: "fabric"
}

explain_fabric_concept

Get detailed explanations of modding concepts with related resources.

// Example: Understand mixins
{
  concept: "mixins"
}

get_minecraft_version

Get current Minecraft version information.

// Get latest version
{ type: "latest" }

// Get all indexed versions
{ type: "all" }

---

Features

Hybrid Search Engine

Combines multiple search strategies for best results:

Strategy	Purpose
FTS5 Full-Text	Fast keyword matching with ranking
Semantic Embeddings	Understanding meaning and context
Section Search	Finding relevant documentation sections
Code Search	Locating specific code patterns

Auto-Updates

The database automatically checks for updates on startup:

• Compares local version with GitHub releases
• Downloads new versions with hash verification
• Creates backups before updating
• Non-blocking - server starts immediately

Documentation Sources

Currently indexes:

• wiki.fabricmc.net - Fabric Wiki (226+ pages)
• docs.fabricmc.net - Official Fabric Docs (266+ pages)

---

For Developers

Development Setup

# Clone repository
git clone https://github.com/OGMatrix/mcmodding-mcp.git
cd mcmodding-mcp

# Install dependencies
npm install

# Run in development mode
npm run dev

Build Commands

# Development
npm run dev              # Watch mode with hot reload
npm run typecheck        # TypeScript type checking
npm run lint             # ESLint
npm run test             # Run tests
npm run format           # Prettier formatting

# Production
npm run build            # Build TypeScript
npm run build:prod       # Build with fresh documentation index
npm run index-docs       # Index documentation with embeddings

Project Structure

mcmodding-mcp/
├── src/
│   ├── index.ts              # MCP server entry point
│   ├── db-versioning.ts      # Auto-update system
│   ├── indexer/
│   │   ├── crawler.ts        # Documentation crawler
│   │   ├── chunker.ts        # Text chunking
│   │   ├── embeddings.ts     # Semantic embeddings
│   │   ├── store.ts          # SQLite database
│   │   └── sitemap.ts        # Sitemap parsing
│   ├── services/
│   │   ├── search-service.ts # Search logic
│   │   └── concept-service.ts # Concept explanations
│   └── tools/
│       ├── searchDocs.ts     # search_fabric_docs handler
│       ├── getExample.ts     # get_example handler
│       └── explainConcept.ts # explain_fabric_concept handler
├── scripts/
│   └── index-docs.ts         # Documentation indexing script
├── data/
│   ├── mcmodding-docs.db     # SQLite database
│   └── manifest.json         # Version manifest
└── dist/                     # Compiled JavaScript

Database Schema

-- Documents: Full documentation pages
CREATE TABLE documents (
  id INTEGER PRIMARY KEY,
  url TEXT UNIQUE NOT NULL,
  title TEXT NOT NULL,
  content TEXT NOT NULL,
  category TEXT NOT NULL,
  loader TEXT NOT NULL,          -- fabric | neoforge | shared
  minecraft_version TEXT,
  hash TEXT NOT NULL             -- For change detection
);

-- Chunks: Searchable content units
CREATE TABLE chunks (
  id TEXT PRIMARY KEY,
  document_id INTEGER NOT NULL,
  chunk_type TEXT NOT NULL,      -- title | section | code | full
  content TEXT NOT NULL,
  section_heading TEXT,
  code_language TEXT,
  word_count INTEGER,
  has_code BOOLEAN
);

-- Embeddings: Semantic search vectors
CREATE TABLE embeddings (
  chunk_id TEXT PRIMARY KEY,
  embedding BLOB NOT NULL,       -- 384-dim Float32Array
  dimension INTEGER NOT NULL,
  model TEXT NOT NULL            -- Xenova/all-MiniLM-L6-v2
);

-- FTS5 indexes for fast text search
CREATE VIRTUAL TABLE documents_fts USING fts5(...);
CREATE VIRTUAL TABLE chunks_fts USING fts5(...);

---

Release Workflow

This project uses release-please for automated releases.

Branch Strategy

Branch	Purpose
dev	Active development
prod	Production releases

How It Works

1. Push commits to dev using conventional commits
2. Release-please maintains a Release PR (dev → prod)
3. When merged, automatic release: npm publish + GitHub release + database upload
4. Changes sync back to dev

See RELEASE_WORKFLOW.md for complete details.

---

Configuration

Environment Variables

Variable	Description	Default
DB_PATH	Custom database path	./data/mcmodding-docs.db
GITHUB_REPO_URL	Custom repo for updates	Auto-detected
MCP_DEBUG	Enable debug logging	false

Disabling Auto-Updates

Set DB_PATH to a custom location to manage updates manually:

DB_PATH=/path/to/my/database.db mcmodding-mcp

---

Statistics

Typical index contains:

Metric	Count
Documentation pages	490+
Searchable sections	2,000+
Code examples	500+
Semantic embeddings	5,000+
Database size	~50 MB

---

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

Quick Contribution Guide

1. Fork the repository
2. Create a feature branch from dev
3. Make changes with conventional commits
4. Submit a PR to dev

---

License

MIT License - see LICENSE for details.

---

Acknowledgments

• Fabric Documentation - Official Fabric documentation
• Fabric Wiki - Community wiki
• Model Context Protocol - MCP specification
• Transformers.js - Local ML embeddings
• better-sqlite3 - Fast SQLite bindings

---