How much can I expect to reduce my token count?

Reductions vary based on initial verbosity, but the tool prioritizes density over raw size; even a 10% reduction is considered a success if it maintains 100% semantic clarity.

Does this skill remove important instructions during compression?

No. The skill uses a lossless verification loop that compares the optimized version against the original to ensure all facts, constraints, and nuances are preserved.

Will it break the structure of my YAML or JSON files?

No, the skill is designed to preserve the structural integrity of data formats while only tightening the string values and descriptions contained within them.

What file types can I optimize with this skill?

It supports Markdown (.md), Plain Text (.txt), YAML (.yaml), and JSON (.json) files, commonly used for system prompts and configuration files.

Is the optimization process fully automated?

Yes, it follows a structured multi-phase workflow including validation, verification, and iterative optimization with a final metrics report.

Prompt Token Optimizer

Name: Prompt Token Optimizer
Author: doodledood

bydoodledood

•

生产力与工作流

Optimizes prompt token efficiency through iterative, lossless compression to reduce consumption while preserving full semantic nuance.

This skill provides an automated, verification-driven workflow for maximizing the information density of Claude prompts, system files like CLAUDE.md, and agent specifications. By employing a multi-pass loop that includes initial verification, targeted optimization, and semantic consistency checks, it significantly reduces token counts without sacrificing complex instructions or stylistic nuances. It is ideal for developers looking to minimize context window usage and lower API costs while maintaining high-performance agent behavior.

主要功能

01Semantic consistency checking to preserve nuance

02Support for Markdown, Text, YAML, and JSON formats

03Automated token reduction through advanced phrasing techniques

04Multi-pass verification loop for lossless compression

058 GitHub stars

06Atomic file replacement after successful verification

使用场景

01Compacting large system instructions or CLAUDE.md files

02Refining agentic prompts to reduce operational costs

03Maximizing information density in developer specifications

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add doodledood/claude-code-plugins optimize-prompt-token-efficiency

For use in Claude.ai and ChatGPT

主要功能

01Semantic consistency checking to preserve nuance

02Support for Markdown, Text, YAML, and JSON formats

03Automated token reduction through advanced phrasing techniques

04Multi-pass verification loop for lossless compression

058 GitHub stars

06Atomic file replacement after successful verification

使用场景

01Compacting large system instructions or CLAUDE.md files

02Refining agentic prompts to reduce operational costs

03Maximizing information density in developer specifications

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add doodledood/claude-code-plugins optimize-prompt-token-efficiency

For use in Claude.ai and ChatGPT