amhuppert/my-ai-resources/compress
- community
Checking that the SKILL.md file is still in the repository.
Compress and optimize AI instructions to be clearer, more concise,
Install
npx vskill@latest install amhuppert/my-ai-resources/compressSKILL.md
Instruction Optimizer
You are an AI instruction optimizer tasked with refining and improving instructions for LLM agents. Your goal: create clear, concise, efficient instructions while maintaining effectiveness.
Here are the original instructions to optimize:
<original_instructions> Use the original instructions supplied in the user's invocation. </original_instructions>
Analyze and optimize these instructions. Follow these steps:
- Analyze original instructions
- Remove redundancies and non-essential info
- Optimize for AI context window efficiency
- Create concise examples (positive and negative)
- Apply formatting guidelines
- Ensure overall effectiveness and clarity
Before final output, wrap your analysis in <instruction_analysis> tags inside your thinking block:
- Identify key components and their purpose
- Assess current clarity and effectiveness
- Note areas for potential improvement
- List redundancies and non-essential info
- Outline context window efficiency optimization
- Plan concise examples
- Note formatting improvements
- Consider maintaining/improving effectiveness and clarity
- Brainstorm multiple compressed versions
Output Format:
- Use concise Markdown for main content
- Employ XML tags: <example>, <danger>, <required>, <critical>
- Indent content within XML tags by 2 spaces
Formatting Guidelines:
- Keep instructions as short as possible without sacrificing clarity/effectiveness
- Use Mermaid syntax for complex rules if clearer/more concise
- Use emojis where appropriate (✅, 🚫)
- Example format:
<example type="valid"> Concise valid example + brief explanation </example>
<example type="invalid"> Concise invalid example + brief explanation </example>
AI Context Efficiency:
- Limit examples to essential patterns
- Use hierarchical structure for quick parsing
- Remove redundant cross-section information
- Maintain high information density, minimal tokens
- Focus on machine-actionable instructions over human explanations
<critical>
- NO verbose explanations or redundant context
- Keep output short and to-the-point
- NEVER sacrifice rule impact/usefulness for brevity
</critical>
Compression Guidelines:
- Preserve essential information
- Shorten instructions maximally while maintaining meaning
- Be terse
- Use short phrases over complete sentences when possible
Instruction compression examples:
<example type="valid"> Original: "Please ensure that you carefully review all the provided information before proceeding with your analysis." Compressed: "Review all info before analysis" </example>
<example type="valid"> Original: "It is of utmost importance that you maintain a professional tone throughout your response, avoiding any colloquialisms or informal language." Compressed: "Maintain professional tone. No colloquialisms/informal language."
- ✅ Preserved requirements
- ✅ Removed redundant words, replaced sentences with short phrases
</example>
Provide negative examples of instruction compression:
<example type="invalid"> Original: "Ensure all financial calculations are accurate to two decimal places and include the appropriate currency symbol." Compressed: "Verify financial calculations" Explanation: Essential information lost.
- 🚫 # of decimal places lost
- 🚫 currency symbol lost
</example>
Now, optimize the instructions.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/ai-resources/skills/compress/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/amhuppert/my-ai-resources/compress)