toon-mcp

Toon MCP Server

toon-mcp helps developers prepare structured data for LLM input using TOON, a format that avoids repeating field names across records. Its tools encode and decode data, estimate token savings, and adapt data-bearing prompts.

Server facts

toon-mcp

Project

Author
elminson
License
MIT
Category
Productivity

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
3
Last commit
Data as of

Tags

  • ai
  • cost reduction
  • data format
  • json
  • llm
  • productivity
  • token optimization

Overview

Repeated keys in tabular records are the focus of toon-mcp, which exposes TOON conversion through MCP. The format places field names in a shared header rather than repeating them throughout a collection. With toon_encode, users can transform JSON, CSV, TSV, XML, HTML tables, YAML, or raw objects into this representation. The reverse workflow uses toon_decode to recover a JSON object or array. These tools serve developers working with API output, database query results, and other structured collections intended for LLM input.

For assessing a conversion before using it, toon_analyze reports potential token savings. For existing prompts, toon_optimize_prompt identifies embedded data structures, rewrites them as TOON, and includes guidance for interpreting the format alongside estimated savings. The documentation describes workflows in Claude Desktop and Claude Code, as well as programmatic use from Node.js. It identifies collections with consistent fields as a stronger fit than deeply nested or irregular data, and notes that small payloads benefit less. The server uses stdio transport and carries an MIT license.

toon-mcp setup notes

  • toon-mcp records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers