How does this skill help with Claude 4.5 specifically?

It implements 4.5-specific behaviors like 'action framing' and 'contextual motivation,' ensuring the model doesn't just suggest changes but actively implements them with higher precision.

How do I know if logic should be a prompt or a script?

The skill includes a 'Logic Extraction' workflow with a 12-point scoring system to help you decide if a task is better suited for deterministic code or LLM reasoning.

Can this skill help reduce my token costs?

Yes, it includes specialized workflows for context optimization that identify static blocks of 1024+ tokens for efficient prompt caching.

Does it support XML tag structures for prompts?

Absolutely. The skill uses XML tag patterns as a core architectural standard to improve Claude's ability to parse complex instructions and data.

Can it fix hallucinations in my current prompts?

The skill provides a guardrail implementation guide that uses quote-grounding, citations, and uncertainty permissions to significantly reduce model hallucinations.

Advanced Prompt Engineering

Name: Advanced Prompt Engineering
Author: thoeltig

bythoeltig

データサイエンスとML

Optimizes and orchestrates advanced prompt engineering workflows for Claude 4.5 using industry-best patterns, guardrails, and context management.

概要

This skill empowers developers to architect high-performance prompts specifically tuned for Claude 4.5's unique reasoning capabilities. It provides structured workflows for analyzing legacy prompts, implementing guardrails to prevent hallucinations, and optimizing context through strategic caching and XML structuring. By integrating core patterns like Chain-of-Thought, action framing, and contextual motivation, the skill ensures that AI-driven features remain consistent, cost-effective, and reliable. Whether you are migrating from older models or building complex agentic workflows, this skill ensures your prompts follow the latest architectural patterns for maximum efficiency.

主な機能

Context caching strategies to reduce token overhead
Logic extraction decision trees (Prompt vs. Script)
0 GitHub stars
Claude 4.5 specific optimization and behavior tuning
Automated prompt analysis and issue categorization
Multi-level guardrail implementation for hallucination prevention

ユースケース

Migrating legacy Claude 3 prompts to Claude 4.5 for better reasoning
Optimizing long-context prompts to leverage 1024+ token caching
Building standardized agentic workflows with explicit success criteria

概要

主な機能

Context caching strategies to reduce token overhead
Logic extraction decision trees (Prompt vs. Script)
0 GitHub stars
Claude 4.5 specific optimization and behavior tuning
Automated prompt analysis and issue categorization
Multi-level guardrail implementation for hallucination prevention

ユースケース

Migrating legacy Claude 3 prompts to Claude 4.5 for better reasoning
Optimizing long-context prompts to leverage 1024+ token caching
Building standardized agentic workflows with explicit success criteria