Terms that get mixed up
Five of these words are used loosely, sometimes for each other. Each one adds something different to the room.
- Tool
- One action the building can carry out for the robot: search, send, look up, calculate. The robot asks for it on a slip.
- Adds reachAlso called a function or an action.
- Connector
- A cable to one app that supplies a bundle of tools, using the open MCP standard.
- Adds a source of toolsA product name. Elsewhere: integration, app, or MCP server.
- Skill
- A binder of written instructions for doing one kind of job well. Opened only when the job calls for it.
- Adds know-howStarted at Anthropic, now an open format. A skill may tell the robot which tools to use, but it is not a tool.
- Memory
- Notes kept in the cabinet between conversations and put back on the desk when relevant.
- Adds continuityNot the same as the desk, which is cleared every time.
- Hook
- A fixed step the building runs at a set moment, whatever the robot thinks about it.
- Adds controlClaude Code’s term. Elsewhere: guardrail, policy, approval step.
In short: tools let the robot act, connectors bring tools in from your apps, skills tell it how to do a job, memory carries notes between conversations, and hooks are steps the building runs whatever the robot decides.
The building has a real name too. It is the software around the model, which runs the tools, the cables and the checkpoint. Engineers call it the harness. Most people call it the app.
Every step, and its real name
| In the room | Real name | What it adds |
| 01 | The room | Language model | Reads text and writes text. Nothing else. |
| 02 | The desk | Context window | Everything the model can see at one time. |
| 03 | A notice on the wall | System prompt | Standing instructions from whoever runs the system. |
| 04 | Scratch paper | Reasoning | Working-out before the answer. |
| 05 | A wider slot | Multimodal input and output | Images, audio and files, not only text. |
| 06 | A phone | Voice | Spoken conversation in both directions. |
| 07 | A filing cabinet | Memory | Notes that survive between conversations. |
| 08 | A bookshelf | Skills | Written procedures, opened only when needed. |
| 09 | A tool hatch | Tool use | A way to ask outside software to act. |
| 10 | Cables to your apps | Connectors, built on MCP | A standard plug that brings an app’s tools with it. |
| 11 | A workshop | Sandbox for running code | A sealed place to run code and operate a computer. |
| 12 | Working in rounds | The agent loop | Permission to keep going, round after round. |
| 13 | A checkpoint | Hooks and permissions | Rules the software enforces every time. |
| 14 | More rooms | Subagents | Helpers with their own robots and their own desks. |
Dates and sources
Dates given to the month were checked against the sources below in November 2026. Where only a year or a general period is given, treat the date as approximate.
- OpenAI: Introducing ChatGPT, 30 November 2022
- Google: Our next-generation model: Gemini 1.5, 15 February 2024
- OpenAI: ChatGPT can now see, hear, and speak, 25 September 2023
- OpenAI: Function calling and other API updates, 13 June 2023
- Anthropic: Introducing the Model Context Protocol, 25 November 2024
- Anthropic: Donating the Model Context Protocol, 9 December 2025, which names ChatGPT, Gemini and Microsoft Copilot among the products that adopted it
- Anthropic: Introducing Agent Skills, 16 October 2025, updated 18 December 2025
Next: The Factory Floor. Pull back, and this room is one workstation on a production line.