ai-prompt-engineering

Prompt Engineering — Operational Skill

Use this skill for production prompt design: schema-first outputs, tool and RAG prompts, prompt hardening, evals, and release workflows. Keep it operational. If the main problem is architecture, retrieval quality, deployment, or inference cost, route to the deeper adjacent skill.

ASCII Flow

prompt job
  |
  v
pattern choice
  structured output | extraction | RAG | tool use | rewrite | classify | release
  |
  v
prompt contract
  inputs + role/task + context rules + output schema + refusal/failure behavior
  |
  v
validation
  schema checks + citation/tool checks + eval cases + regression gate
  |
  v
released prompt
  versioned artifact + rollout notes + rollback path

When to Use This Skill

• designing or refactoring prompts for production LLM systems
• structured outputs, extraction schemas, or response contracts
• prompt debugging, prompt hardening, or prompt review
• prompt evals, regression suites, and rollout criteria
• tool-use or RAG prompt patterns
• multimodal prompts for image, document, audio, or video inputs

Route Elsewhere

• agent architecture and orchestration -> ai-agents
• retrieval quality and chunking -> ai-rag
• broader LLM lifecycle and model strategy -> ai-llm
• inference latency and cost optimization -> ai-llm-inference
• deployment, monitoring, and platform controls -> ai-mlops

---

Quick Start

1. Classify the prompt job: structured output, extraction, RAG, tool use, rewrite, classification, or release workflow.
2. Start from a template or provider-native prompt feature rather than writing from scratch.
3. Add explicit output and refusal rules.
4. Add validation: schema checks, citation checks, post-tool checks, and failure handling.
5. Add evals before calling the prompt production-ready.

Quick Reference

• Pattern selection -> ## Pattern Chooser
• Reusable prompt shapes -> ## Minimal Prompt Skeletons
• Release hardening -> ## Production Checklist
• Deeper references and templates -> ## Navigation