Glossary
The vibe coder's dictionary
Every term we use, defined in plain English.
A
Agent
An LLM-powered process that can take multi-step actions with tools.
Anchor prompt
A base prompt reused across variations to keep tone and constraints consistent.
C
Chain-of-thought
A prompting technique that asks the model to reason step-by-step.
Context window
The maximum number of tokens a model can consider at once.
E
Embedding
A numeric vector representing the meaning of text, used for semantic search.
F
Fine-tuning
Training a base model on additional examples to shape its behavior.
Focus session
A timed, distraction-free block of work the core loop in Build on Vibe.
G
Grounding
Constraining an AI's output with authoritative source material.
H
Hallucination
When a model generates plausible-sounding but false content.
L
LLM
Large language model the class of models that power vibe coding tools.
P
Prompt pack
A curated bundle of prompts for a specific outcome.
R
RAG
Retrieval-augmented generation grounding output on retrieved documents.
S
Spec-first
Writing a short specification before prompting, to reduce rework.
System prompt
Persistent instructions given to a model before any user input.
T
Temperature
A dial that controls how creative or deterministic an output is.
Token
The unit of text an LLM consumes; roughly ¾ of a word in English.
V
Vibe coding
Building software primarily by describing intent to an AI.