How does it optimize for the Claude context window?

It identifies 'Lost-in-the-Middle' issues where critical instructions are buried in the center of a long prompt, recommending they be moved to the beginning or end where the model's attention is highest.

Can the Prompt Optimizer automatically fix my prompt files?

Yes, by using the --fix flag, the skill will automatically apply corrections to high-certainty issues, such as converting negative-only constraints into positive instructions.

What is the difference between enhance-prompts and enhance-agent-prompts?

The enhance-prompts skill focuses on general prompt quality such as clarity and structure, while enhance-agent-prompts is specifically designed for agent configuration files containing YAML frontmatter.

Why does this skill flag 'think step-by-step' as an anti-pattern?

Modern models like Claude 3.5 and 4.x are fine-tuned for internal reasoning. Explicitly asking them to think step-by-step is often redundant and consumes unnecessary tokens without significantly improving accuracy.

Prompt Optimizer

Name: Prompt Optimizer
Author: avifenesh

byavifenesh

•

311

•

生产力与工作流

Improves AI prompts by analyzing structure, clarity, and reliability based on Claude-specific best practices.

The Prompt Optimizer skill acts as an automated prompt engineer for Claude Code, scanning your markdown or text files to identify clarity issues, structural weaknesses, and anti-patterns. It leverages deep knowledge of Claude’s architecture—including XML tagging, context window behavior, and modern reasoning capabilities—to transform vague or inefficient instructions into high-performing system and user prompts. By applying research-backed patterns like few-shot examples and instruction hierarchy, it ensures your AI interactions are deterministic, reliable, and token-efficient.

主要功能

01Automated analysis of prompt clarity, structure, and output reliability

02Context window optimization to prevent 'Lost-in-the-Middle' performance degradation

03Auto-fix capabilities for high-certainty issues like aggressive emphasis and negative constraints

04Claude-specific XML tag optimization and few-shot example validation

05Detection and removal of redundant Chain-of-Thought instructions

06311 GitHub stars

使用场景

01Refining complex system prompts for AI agents to ensure deterministic output

02Standardizing prompt templates across a repository to improve team collaboration

03Modernizing legacy prompts for Claude 3.5 and 4.x by removing redundant reasoning steps

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add avifenesh/awesome-slash prompts

For use in Claude.ai and ChatGPT

Download Skill