moltlang-mcp-server
v0.1.0
Published
MoltLang MCP Server - AI-optimized language for efficient agent-to-agent communication
Downloads
48
Maintainers
Readme
MoltLang MCP Server
Model Context Protocol (MCP) server for MoltLang - an AI-optimized language that reduces token consumption by 50-70% for efficient agent-to-agent communication.
Quick Start
Installation
npm install
npm run buildUsage
# Start server (stdio transport - default for MCP clients)
npm run start:stdio
# Or start with HTTP transport
npm startMCP Client Configuration
Add to your MCP client configuration (e.g., Claude Desktop):
{
"mcpServers": {
"moltlang": {
"command": "node",
"args": ["/path/to/mcp-server/dist/index.js"]
}
}
}Available Tools
| Tool | Description | Token Cost |
|------|-------------|------------|
| molt | English→MoltLang (50-70% reduction) | ~500 tokens |
| unmolt | MoltLang→English | ~400 tokens |
| validate_molt | Validate translation quality | ~450 tokens |
| list_tokens | List available MoltLang tokens | ~300 tokens |
| get_efficiency | Calculate token efficiency | ~350 tokens |
Architecture
┌─────────────────────────────────────────────────────────┐
│ MCP Client (Claude, etc.) │
└────────────────────┬────────────────────────────────────┘
│
│ MCP Protocol (JSON-RPC)
▼
┌─────────────────────────────────────────────────────────┐
│ TypeScript MCP Server │
│ ┌──────────────────────────────────────────────┐ │
│ | Tool Definitions (minimal descriptions) | │
│ └──────────────────────────────────────────────┘ │
│ ┌──────────────────────────────────────────────┐ │
│ | Translation Handlers | │
│ └──────────────┬───────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌──────────────────────────────────────────────┐ │
│ | Python Bridge (spawn subprocess) | │
│ └──────────────┬───────────────────────────────┘ │
└─────────────────┼───────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────┐
│ Python MoltLang Package │
│ (translator.py, tokens.py, validator.py) │
└─────────────────────────────────────────────────────────┘Development
# Watch mode for development
npm run dev
# Run tests
npm test
# Lint code
npm run lint
# Format code
npm run formatToken Efficiency
This server is designed to minimize MCP token overhead:
- Concise tool descriptions - <50 chars per tool
- Minimal parameter schemas - Only required fields
- Compact responses - Abbreviated field names
- Caching - Common translations cached locally
License
AGPL-3.0 - See LICENSE in parent directory
