Standardizes agent behavior and tool documentation using research-backed XML structures and high-impact prompt engineering techniques.
This skill provides a comprehensive framework for building high-performance AI agents through empirical prompt engineering. It combines a rigorous XML tag hierarchy with structural templates for tool documentation and agent definitions, ensuring consistent enforcement of safety constraints and operational procedures. By incorporating measured interventions like persistence framing and strategic context positioning, it achieves 15-30% improvements in task success rates while minimizing token bloat through direct, imperative communication patterns and the avoidance of common anti-patterns.
主要功能
01High-impact interventions including persistence and tool verification
02Strict XML tag hierarchy for multi-level instruction enforcement
030 GitHub stars
04Research-backed structural templates for agents and tool documentation
05Empirically-validated writing style and voice guidelines
06Optimization for long-context performance via strategic positioning
使用场景
01Optimizing existing system prompts to improve task success rates by up to 30%
02Documenting specialized tools to improve LLM invocation accuracy and reduce hallucinations
03Designing complex AI agents with specific roles and inviolable safety constraints
