# Glossary and ID Guide

Short definitions of the terms the SDK docs use, and a table of which ID goes where.

## Terms

| Term | Meaning |
|---|---|
| Agent | The deployed actor. It combines one intellect with one or more avatars. Its id is `agentId` |
| Intellect | An agent's configuration: prompts, tools, capabilities and knowledge. Its id is `configId`. The API calls the same object a config |
| Brain | Informal word for what the intellect drives at runtime: the model that decides what the agent says and does. Docs say "agent" for the runtime and "intellect" for the configuration you edit. The word survives in API names such as `brainStalled` and `brainSegment` |
| Avatar | The face and voice half of an agent. Its id is `avatarId` |
| Widget | A public, secret-free handle to one agent. Safe to ship in browser code. Its id is `widgetId` |
| Channel | One of the two live WebRTC connections of an avatar session: the microphone channel going up and the video channel coming down. See [Audio Channels](https://kaltura.github.io/intelligent-agents-sdk/reference/wire-protocol/audio-channels/index.md) |
| Session | One live conversation between a user and an agent. The SDK has no separate "room" object. A session owns its channels |
| Thread | The stored history of a conversation. Its id is `threadId` |
| KS | Kaltura Session. The signed string that proves who is calling and what they may do. Every API call carries one. How to mint one: [Authentication](https://kaltura.github.io/intelligent-agents-sdk/reference/api/authentication/index.md#authentication) |
| Request variables | Values your app sends to the agent, read in prompts as `{{var}}`. See [Dynamic data injection](https://kaltura.github.io/intelligent-agents-sdk/guides/dynamic-data-injection/index.md) |

## Which ID where

| ID | Identifies | You get it from | You use it for |
|---|---|---|---|
| `configId` | The intellect | `provision()`, or `intellects.create` | `converse()` and `converseOnce()`, `sessions.createConversationToken`, `intellectConfig` calls, `setForcedLanguage` |
| `agentId` | The agent | `provision()`, or `agents.create` | `sessions.createAgentToken`, `application.resolveWidgetId`, `agents.update`, labeling threads so [lifecycle rules](https://kaltura.github.io/intelligent-agents-sdk/reference/lifecycle/index.md#scoping-a-rule-to-one-agent) can match them, `setForcedLanguage` |
| `avatarId` | The avatar (a 24-character hex string) | `provision()`, or `avatars.create` | `avatarIds` when you create an agent |
| `widgetId` | The public widget of one agent | `provision()`, or `application.resolveWidgetId(agentId, ks)` | `sessions.createWidgetToken({ widgetId })` for a browser embed with no secret |

`provision()` returns all four. Keep `configId` and `agentId` on your server. Hand the browser a token, or a `widgetId`, never an admin secret.