What are the emphasis levels in prompt-creation?

The skill uses a 6-level syntax ranging from Level 1 (suggestions like 'Please do') to Level 6 (absolute rules like 'CRITICAL - ZERO TOLERANCE') to signal the importance of specific constraints.

How does 'Precognition' improve prompt performance?

Precognition uses tags to force the model to reason through its logic and analyze key points before generating a final answer, which significantly reduces hallucinations.

What is the 9-element prompt structure used by this skill?

It is a comprehensive framework covering Task Context, Tone, Rules, Examples (Few-shot), Input Data, Immediate Task, Precognition (Scratchpad), Output Formatting, and Prefilling.

Can I use these templates for models other than Claude?

While specifically optimized for Anthropic's models and conventions, the structured logic, XML tagging, and few-shot patterns are highly effective across most modern LLMs.

Claude Prompt Engineer

Name: Claude Prompt Engineer
Author: fusengine

byfusengine

•

Data Science & ML

Generates high-performance, structured prompts using official Anthropic conventions and 2025 best practices.

This skill equips Claude Code with a standardized framework for architecting sophisticated prompts based on Anthropic’s official 9-element structure. It provides an engineering-first approach to prompt design, incorporating advanced techniques like Chain-of-Thought (precognition) via scratchpad tags, few-shot examples, and a 6-level hierarchical emphasis system. By following a rigorous creation workflow, it helps developers build robust system instructions that minimize ambiguity, prevent hallucinations, and optimize cognitive performance for complex agent systems.

Key Features

01Implements the official Anthropic 9-element prompt structure

02Automates prompt refinement from objective definition to edge-case testing

03Uses 'Precognition' scratchpad tags for enhanced logical reasoning

041 GitHub stars

05Applies a 6-level hierarchy for rule enforcement and constraint weighting

06Standardizes few-shot examples using XML-tagging conventions

Use Cases

01Architecting complex system prompts for autonomous AI agents

02Converting informal developer instructions into structured production-grade prompts

03Refining existing LLM instructions to improve reliability and output formatting

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fusengine/agents prompt-creation

For use in Claude.ai and ChatGPT

Download Skill