Vibe Coding Glossary

Essential Terms

Agentic Coding

AI systems that autonomously plan, execute, and verify multi-step coding tasks — reading files, running tests, debugging errors, and iterating without human intervention per step.

Chain-of-Thought (CoT)

A prompting technique that asks the AI to break down its reasoning into steps before generating a final answer. Produces more accurate and debuggable code.

Context Window

The maximum amount of text (measured in tokens) an AI model can process in a single interaction — including system prompts, file contents, and conversation history.

Cursor Rules (.cursorrules)

A project-specific configuration file that instructs the Cursor AI IDE about coding conventions, architectural patterns, and project-specific requirements.

Hallucination

When an AI generates plausible-looking code that references APIs, methods, or patterns that don't actually exist. Common with newer or less-documented libraries.

LLM (Large Language Model)

The neural networks (GPT-4, Claude, Gemini) that power AI coding assistants. They predict the most likely next tokens based on patterns learned from training data.

MCP (Model Context Protocol)

A standard protocol for connecting AI assistants to external tools and data sources — databases, APIs, documentation. Think of it as the USB standard for AI tooling.

Prompt Engineering

The practice of crafting precise instructions to AI systems to maximize output quality. In code, this means specifying language, constraints, patterns, and expected output format.

RAG (Retrieval-Augmented Generation)

A technique where AI retrieves relevant information from external sources (documentation, codebases) before generating responses, improving accuracy for domain-specific questions.

Token

The fundamental unit of text that AI models process — roughly 4 characters or 0.75 words in English. Context windows, costs, and speed are all measured in tokens.

Vibe Coding

A development approach where developers describe intent in natural language and let AI handle implementation, focusing on architecture and review rather than syntax. Term coined by Andrej Karpathy.