이 용어집은 프롬프트 엔지니어링에서 가장 중요한 500개 용어를 다룹니다. 기초 개념부터 에이전트 오케스트레이션과 평가 프레임워크까지 망라합니다. 각 항목은 개발자와 AI 실무자를 위해 간결하고 실용적인 정의를 제공하며, 심화 학습을 위한 기본 참고 링크도 포함합니다.
용어는 여섯 가지 그룹으로 구성됩니다: 핵심 프롬프팅 개념, 에이전트 및 오케스트레이션, 안전성 및 정렬, 평가 및 테스트, 고급 기법, 지표 및 프로덕션. 검색 가능한 표를 빠른 참조로 활용하거나 링크를 따라가 구현 세부 사항을 확인하시기 바랍니다.
핵심 프롬프팅 개념
Prompt
Prompt engineering
PromptingGuide Overview, LearnPrompting Definition, IBM Techniques
LLM (Large Language Model)
Context window
Wikipedia, Firecrawl Context Engineering, PromptingGuide Settings
System prompt
Hallucination
Grounding
Zero-shot prompting
PromptingGuide Zero-shot, Codecademy Shot Prompting, Lakera 2026
Few-shot prompting
Chain-of-Thought (CoT)
Zero-shot CoT
Role prompting
LearnPrompting Roles, PromptingGuide Basics, DecodeTheFuture 2026
Prompt chaining
Anthropic Chain Prompts, PromptingGuide Chaining, Lakera Orchestration
Tree-of-Thought (ToT)
PromptingGuide ToT, LearnPrompting Tree of Thought, ClipboardAI Glossary
Temperature
PromptingGuide Settings, Tetrate Guide, PromptEngineering.org
Top-p (nucleus sampling)
PromptEngineering.org Temperature & Top-p, PromptingGuide Settings, Infomineo Best Practices
RAG (Retrieval-Augmented Generation)
Open Weights
Meta – LLaMA Community License, Mistral AI – License, Wikipedia – Open-weights models
Fine-tuning
Anthropic – Fine-tuning guide, OpenAI – Fine-tuning API, IBM – RAG vs fine-tuning
LoRA
Hu et al. – LoRA paper, Dettmers et al. – QLoRA paper, PromptingGuide – Advanced techniques
VRAM
NVIDIA – GPU memory, Ollama – Hardware guide, HuggingFace – Model cards
Context engineering
에이전트 및 오케스트레이션
Agent
OpenAI Agents – Orchestration, Genesys – LLM agent orchestration, GetStream – AI agent orchestration
Tool
IBM – What is tool calling?, LLMBase – Tool call, OpenAI – Tools & function calling
Tool call
Tool schema
OpenAI – Tool specification, IBM – Tool calling guide, OpenAI Agents SDK
Agent orchestration
OpenAI – Agent orchestration, Genesys – LLM agent orchestration, IBM – Orchestration tutorial
Multi-agent system
Eonsr – Orchestration frameworks 2025, Zylos – Multi-agent patterns 2025, GetStream – AI agent orchestration
Planner agent
OpenAI Agents – Planning, IBM – Orchestration tutorial, Zylos – Multi-agent patterns
Executor agent
OpenAI Agents SDK, Genesys – Agent orchestration, GetStream – Orchestration
Router agent
OpenAI – Routing patterns, Eonsr – Orchestration frameworks, Zylos – Multi-agent patterns
Guardrail
Lakera – Prompt engineering & safety, Zendesk – AI glossary (guardrails), GetStream – Orchestration best practices
Observation
IBM – Tool calling, OpenAI Agents – Tools, Genesys – Orchestration flows
State (agent state)
OpenAI – Agent orchestration, IBM – Orchestration tutorial, Zylos – Production considerations
Memory (short-term)
PromptingGuide – Context & history, OpenAI – Conversation design, CoherePath – Glossary
Memory (long-term)
Firecrawl – Context engineering, Zylos – Multi-agent production, PromptingGuide – RAG & memory
Vector store
PromptingGuide – RAG, AWS – Vector databases overview, Eonsr – Orchestration frameworks
Action space
OpenAI Agents – Actions & tools, IBM – Agent orchestration guide, GetStream – Orchestration best practices
Termination condition
OpenAI – Agent orchestration, Zylos – Production considerations, Multi-agent patterns video
Sequential orchestration
Multi-agent patterns video, OpenAI – Orchestration patterns, Genesys – Orchestration
Parallel orchestration
Zylos – Multi-agent orchestration 2025, Multi-agent patterns video, Eonsr – Orchestration frameworks
Producer-reviewer loop
Multi-agent patterns video, GetStream – Orchestration, IBM – Orchestration tutorial
안전성 및 정렬
Safety policy
OpenAI – Safety best practices, Anthropic – Safety overview, Lakera – Safety & guardrails
Guardrails
Anthropic – Safety & guardrails, OpenAI – Safety best practices, Zendesk – Generative AI glossary
Prompt injection
OWASP – LLM prompt injection, Lakera – Prompt injection, Microsoft – Prompt injection guidance
Jailbreak
OWASP – LLM jailbreaks, Lakera – Jailbreak examples, Anthropic – Safety FAQ
Red-teaming
Anthropic – Red-teaming AI systems, OpenAI – Safety & red teaming, OWASP – Testing LLM apps
Toxicity
Google – Perspective API, Zendesk – AI glossary, OpenAI – Safety best practices
Bias
OpenAI – Addressing bias, IBM – Bias in AI, Anthropic – Responsible scaling
Alignment
Anthropic – Constitutional AI, OpenAI – Alignment & safety, DeepMind – Alignment research
RLHF
OpenAI – RLHF paper, Anthropic – RL from AI feedback, DeepMind – RLHF overview
Constitutional AI
Anthropic – Constitutional AI, Anthropic – Research paper, Zendesk – AI glossary
평가 및 테스트
Evals (evaluation suite)
OpenAI – Evals framework, Anthropic – Model evaluations, ClipboardAI – AI glossary
Golden set
OpenAI – Evals docs, Microsoft – Evaluation guidance, Anthropic – Evaluating Claude
A/B prompt test
OpenAI – Prompt best practices, KeepMyPrompts – Testing prompts, Lakera – Prompt optimization
Win rate
OpenAI – Evals & comparison, Anthropic – Model evals, Microsoft – Evaluation patterns
Regression test
OpenAI – Evals, Microsoft – Regression evaluation, OWASP – LLM application testing
Human-in-the-loop (HITL)
Microsoft – Responsible AI, OpenAI – Safety best practices, Anthropic – Human feedback
Monitoring
Datadog – LLM observability posts, Microsoft – Monitoring guidance, OWASP – LLM security
Drift
Google – ML data drift, OpenAI – Monitoring, Eonsr – Orchestration in production
Prompt versioning
KeepMyPrompts – Prompt management, Lakera – Prompt lifecycle, OpenAI – Prompting best practices
Prompt repository
OpenAI – Prompt library examples, CoherePath – Prompting glossary, ClipboardAI – AI glossary
고급 기법
Self-Consistency
PromptingGuide – Self-Consistency, IBM – Prompt techniques, Lakera – Prompt engineering guide
Meta-Prompting
PromptingGuide – Meta Prompting, IBM – Prompt engineering techniques, DigitalApplied – Advanced techniques 2026
Automatic Prompt Engineer (APE)
PromptingGuide – Automatic Prompt Engineer, PromptingGuide – Techniques, K2View – Prompt techniques 2026
Reflexion
PromptingGuide – Reflexion, PromptingGuide – LLM Agents, Lakera – Advanced guide
Multimodal Prompting
Promptitude – Prompt engineering 2026, PromptingGuide – Multimodal CoT, Promnest – Best practices 2026
Graph-of-Thoughts (GoT)
PromptingGuide – Techniques, Promnest – Cognitive architectures 2026
Chain-of-Table
GetMaxim – Advanced techniques 2025/2026, PromptingGuide – Advanced techniques
Active-Prompt
Directional Stimulus Prompting
PromptingGuide – Directional Stimulus Prompting, PromptingGuide – Techniques overview
프로그램 보조 언어 모델 (PAL)
PromptingGuide – Program-Aided Language Models, PromptingGuide – Advanced
Agentic RAG
LinkedIn – Agentic AI terms, K2View – Agentic RAG, Reddit – Agentic terms
Handoff (agent handoff)
OpenAI Agents SDK – Handoffs, Zylos – Multi-agent patterns, Genesys – Orchestration
Orchestrator agent
OpenAI – Agent orchestration, Eonsr – Orchestration frameworks 2025, Zignuts – Prompt engineering guide
Critic / Reviewer agent
Multi-agent patterns, IBM – Orchestration tutorial, GetStream – Best practices
GraphRAG
Prompt Tuning
Zendesk – Generative AI glossary, IBM – RAG vs fine-tuning vs prompting
Context Compression
Adaptive Prompting
Promptitude – Trends 2026, RefonteLearning – Optimizing interactions 2026
G-Eval
Microsoft – Evaluation guidance, Confident AI – LLM evaluation metrics
지표 및 프로덕션
BERTScore
ROUGE
BLEU
Perplexity
Answer Relevancy
Task Completion Rate
Prompt Injection (indirect)
OWASP – LLM top 10, Penligent – Agent hacking 2026, Microsoft – Guidance
Agent Hijacking
인간 검토 포함 평가 (HITL Evaluation)
LLM-as-a-Judge
Prompt Repository (enterprise)
OpenAI – Examples, Braintrust – Prompt tools 2026, KeepMyPrompts – Management
Prompt Optimizer
Dev.to – Automatic prompt optimization, Braintrust – Tools 2026
Multi-Modal Orchestration
Shadow AI
Constitutional AI (extended)
Drift Detection (prompt/model)
Google – ML drift, Eonsr – Production, Datadog – Observability
Win Rate (pairwise)
OpenAI – Evals, Anthropic – Model evaluations, Microsoft – Evaluation
Context Engineering (advanced)
Firecrawl – Context engineering, AIPromptLibrary – Advanced 2026, KeepMyPrompts – Guide
Swarm / Collective Intelligence
Zignuts – Prompt engineering guide, Promnest – Orchestration
Prompt Versioning & Rollback
KeepMyPrompts – Prompt management, Lakera – Prompt lifecycle, Braintrust – Tools
자주 묻는 질문
프롬프트 엔지니어링이란 무엇입니까?
프롬프트 엔지니어링은 언어 모델이 유용하고 예측 가능하며 안전한 출력을 생성하도록 프롬프트를 설계하고 반복하는 분야입니다. 신뢰성과 품질을 향상시키기 위해 지침을 구조화하고, 맥락을 추가하며, few-shot 또는 chain-of-thought와 같은 기법을 선택하는 작업이 포함됩니다.
Zero-shot 프롬프팅과 few-shot 프롬프팅의 차이점은 무엇입니까?
Zero-shot 프롬프팅은 예시 없이 지침만으로 모델에게 과제를 수행하도록 요청하는 방식으로, 모델의 사전 훈련이 이미 해당 패턴을 다루는 일반적인 과제에 가장 적합합니다. Few-shot 프롬프팅은 실제 쿼리를 처리하기 전에 모델이 원하는 패턴, 형식 또는 스타일을 추론할 수 있도록 프롬프트에 소수의 입출력 예시를 포함합니다. Few-shot 방식은 복잡하거나 일반적이지 않은 과제에서 일반적으로 더 높은 품질을 생성합니다.
AI에서 RAG는 무엇을 의미합니까?
RAG는 검색 증강 생성(Retrieval-Augmented Generation)의 약자입니다. 훈련 데이터에만 의존하는 것이 아니라 현재의 근거 있는 데이터를 기반으로 모델이 답변하도록 지식 기반에서 관련 문서를 검색하여 프롬프트에 삽입하는 아키텍처입니다. 이를 통해 환각을 줄이고 답변이 실제적이고 최신 정보를 기반으로 하도록 보장합니다.
프롬프트 엔지니어링과 파인튜닝의 차이점은 무엇입니까?
프롬프트 엔지니어링은 모델 자체를 변경하지 않고 프롬프트를 설계하고 반복하여 모델 출력을 유도하는 분야입니다. 반면 파인튜닝은 과제별 데이터로 훈련하여 모델의 가중치를 수정합니다. 프롬프트 엔지니어링은 더 빠르고, 저렴하며, 반복하기 쉽습니다. 파인튜닝은 전문화된 과제에서 더 나은 결과를 달성할 수 있지만 더 많은 데이터와 계산 자원이 필요합니다.
AI에서 컨텍스트 윈도우란 무엇입니까?
컨텍스트 윈도우는 시스템 프롬프트, 대화 기록, 검색된 문서를 포함하여 모델이 한 번에 고려할 수 있는 최대 토큰 수입니다. 컨텍스트 제한을 초과하면 오래된 또는 중간 부분의 컨텍스트가 잘리거나 무시됩니다. 더 긴 컨텍스트는 더 비싸고 처리 속도가 느리기 때문에 비용과 지연 시간을 관리하는 데 컨텍스트 윈도우 크기를 이해하는 것이 중요합니다.
