Glossary

Every AI term clearly explained, from agent to VPS.

A

Agent: An AI that doesn't just answer but takes steps itself: uses tools, reads files, performs actions, in a loop until the task is done.
AGENTS.md: An instruction file in a project telling coding agents how the project works. Now supported by nearly all coding agents.
API: A way for programs to talk to each other. Through an AI provider's API you use a model from your own code or tools, paying per token.

C

Context window: How much text (in tokens) a model can 'see' at once: your question, the conversation, documents and tool results combined.

H

Hallucination: When an AI confidently makes up something that isn't true. Always verify facts, numbers and sources.

L

LLM (Large Language Model): Software trained on huge amounts of text that predicts which piece of text logically comes next.
Local model: An AI model running on your own computer instead of the cloud. Private and offline, but limited by your hardware.

M

MCP (Model Context Protocol): An open standard for connecting AI apps to tools and data (like GitHub, your files or a database) via 'MCP servers'. Governed by the Agentic AI Foundation.

O

Open weights: A model whose 'weights' (the trained model itself) can be downloaded so you can run it yourself. Mind the licence.

P

Prompt: The instruction or question you give an AI.
Prompt injection: An attack where hidden instructions in text (a web page, email, document) make an AI agent do what the attacker wants.

Q

Quantisation: Making a model smaller by storing numbers less precisely (e.g. Q4) so it fits in less memory — at a small quality cost.

R

RAG (Retrieval-Augmented Generation): Having the AI first look up relevant parts of your documents and pass them along with the question, so the answer is grounded in your data.
Reasoning model: A model that 'thinks' first (generates intermediate steps) before answering. Better at hard tasks, but slower and pricier.

S

Skill (Agent Skill): A folder with a SKILL.md file (plus optional scripts) that teaches an agent how to perform a specific task. Open standard, supported by dozens of tools.
SSH: An encrypted connection to log into a remote server and run commands. Preferably with a key pair instead of a password.See the lesson

T

Token: The chunk of text a model works in: roughly three quarters of a word in English, often less in Dutch. You pay per token.

V

VPS (Virtual Private Server): A rented virtual server in a data centre, always on, for a few euros a month. Ideal for a 24/7 agent.