Token-Oriented Object Notation (TOON) is a compact data format specifically designed to maximize LLM context efficiency while maintaining structural integrity and readability for AI agents.

Does this require a specific library to work?

While Claude can understand the notation natively via this skill, the 'toon-format' Python library is available for developers who want to encode or decode TOON data programmatically.

Can I use TOON for external API calls?

No, TOON is optimized for internal agent-to-agent communication and prompting. For external API integrations or instances where standard JSON.parse() is required, you should continue using JSON.

How much context can I save using this skill?

TOON typically uses 30-60% fewer tokens than equivalent JSON structures, which allows for more reasoning space, longer history, or larger inputs within the LLM's context window.

Why use schema.org with TOON?

Using schema.org provides a globally recognized, standardized vocabulary (like SearchAction or ItemList), ensuring that different agents interpret tasks and data objects with zero ambiguity.

TOON Agent Protocol

Name: TOON Agent Protocol
Author: rayk

byrayk

•

생산성 및 워크플로

Optimizes agent-to-agent communication and task delegation using token-efficient Token-Oriented Object Notation (TOON) and schema.org vocabulary.

소개

This skill implements TOON (Token-Oriented Object Notation) combined with schema.org semantics to streamline how Claude interacts with subagents and structures complex tasks. By using a compact syntax that is 30-60% more token-efficient than JSON, it preserves valuable context window space while providing high semantic clarity through standardized types like SearchAction and CreateAction. It is ideal for developers building multi-agent workflows, orchestrating background tasks, or requiring highly structured, consistent responses without the overhead of traditional data formats.

주요 기능

Token-efficient TOON syntax reducing overhead by up to 60% compared to JSON
Structured Action-based prompting patterns for Search, Create, Update, and Assess tasks
Mandatory metadata support for @type and @id to ensure semantic interoperability
Tabular array notation for highly efficient handling of uniform data sets
Standardized schema.org vocabulary for unambiguous task delegation and execution
1 GitHub stars

사용 사례

Delegating complex code analysis or refactoring tasks to subagents with strict output requirements
Reducing context window consumption in long-running agentic conversations and multi-step workflows
Standardizing internal data passing and response formats between different LLM components

소개

주요 기능

Token-efficient TOON syntax reducing overhead by up to 60% compared to JSON
Structured Action-based prompting patterns for Search, Create, Update, and Assess tasks
Mandatory metadata support for @type and @id to ensure semantic interoperability
Tabular array notation for highly efficient handling of uniform data sets
Standardized schema.org vocabulary for unambiguous task delegation and execution
1 GitHub stars

사용 사례

Delegating complex code analysis or refactoring tasks to subagents with strict output requirements
Reducing context window consumption in long-running agentic conversations and multi-step workflows
Standardizing internal data passing and response formats between different LLM components