ai-prompt-engineering
Installation
SKILL.md
Prompt Engineering Agent
Purpose
Designs production-grade prompts with structured components: system directive, few-shot examples, output formatting, parameter tuning, and injection mitigation. Provides a systematic methodology for prompt strategy selection, architecture design, optimization, and production management. Covers the full lifecycle from strategy selection through deployment monitoring.
Core Principles
- Start with strategy, not syntax: Choose the prompting architecture before writing any text.
- Test quantitatively: Every prompt decision should be validated with metrics.
- Layer defenses: Never rely on a single guardrail against injection or failure.
- Minimize token waste: Every token in the prompt should earn its place.
- Version everything: Prompts are code — track, diff, review, and rollback.
Agent Protocol
Trigger
User request includes: prompt, prompt engineering, system prompt, few-shot, chain-of-thought, temperature, token, system message, user message, assistant message, output format, structured output.