Can I use this for non-technical content?

While designed for skills and agents, its core principles of brevity and clarity apply to any markdown-based content intended for consumption by an LLM.

What is the primary goal of the LLM Content Optimizer skill?

It ensures LLM-targeted content remains concise and effective by favoring the correction and clarification of existing instructions over the addition of new ones.

Why is correcting instructions better than adding new ones?

Undesired LLM behavior often stems from incorrect existing information. Adding more instructions increases complexity and the risk of conflict without addressing the root cause.

How does it handle complex technical documentation?

It advocates for progressive disclosure, suggesting that detailed content be moved to a dedicated reference directory while keeping core instructions focused.

When does this skill automatically activate?

The skill triggers when you edit SKILL.md files, modify agent or command markdown files, or specifically ask if content is getting too long or bloated.

LLM Content Optimizer

Name: LLM Content Optimizer
Author: it-bens

byit-bens

0•

内容管理

Refines and optimizes LLM-targeted instructions by prioritizing content correction over additive bloat.

The LLM Content Optimizer skill is designed to maintain the efficiency and clarity of instructions, agents, and command markdown files within your development environment. By enforcing a 'correction over addition' philosophy, it helps developers avoid instruction bloat and prompt fatigue that can lead to degraded model performance. It provides a structured decision framework to determine if new instructions are truly necessary or if existing prompts can be clarified, ensuring your AI-targeted content remains concise, orthogonal, and high-impact.

主要功能

01Progressive Disclosure Support: Recommends moving granular technical details to reference files.

02Conflict Detection: Evaluates proposed changes for potential contradictions with existing guidance.

03Instruction De-bloating: Actively identifies and prevents redundant additions to prompt files.

04Automated Triggers: Activates automatically during SKILL.md, agent, and command markdown edits.

050 GitHub stars

06Correction Framework: Guides users to fix existing logic errors instead of stacking new rules.

使用场景

01Streamlining project documentation to ensure high-density, high-impact instructions for AI models.

02Optimizing Claude Code skills to reduce token usage and improve instruction following.

03Auditing AI agent instructions when behavior becomes erratic due to prompt saturation.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add it-bens/claude-code-plugins content-editing

For use in Claude.ai and ChatGPT

Download Skill