BrokerBot gives you two ways to integrate: a chat widget you embed in your site, and a REST API your backend calls with a team API key.
Start here
- Authentication: get an API key, how keys are scoped to your team, and the base URLs.
- Errors: the error format and what each status code means.
Chat Widget
- Quick Start: embed the widget, sign members in with server-side SSO, pass page context, open a document in chat, and control the widget from your own page (including headless mode).
- SSO Tokens:
POST /auth/generate-tokenissues a one-time token that signs a member in to the widget.
API reference
| Endpoint | Description |
|---|---|
POST /v1/knowledge/search |
Search a team’s knowledge base. |
GET /v1/teams/members |
List a team’s members. |
POST /v1/teams/members |
Add members to a team. |
DELETE /v1/teams/members |
Remove a member from a team. |
GET /v1/teams/groups |
List a team’s groups. |
GET /v1/teams/groups/{groupId}/members |
List a group’s members. |
GET /v1/training |
List a team’s training. |
GET /v1/training/{trainingId} |
Get one training record. |
POST /v1/training/{trainingId}/approve |
Approve training so the agent uses it. |
POST /v1/training/{trainingId}/reject |
Reject training. |
GET /v1/workflows |
List a team’s workflows. |
GET /v1/workflows/{workflowId} |
Get one workflow with its instructions. |
Every /v1/teams endpoint also has a form that takes a sub-team in the path, for example GET /v1/teams/{teamId}/members. See Teams.
Libraries
brokerboton npm: the official TypeScript/JavaScript client. It covers knowledge search and SSO tokens. Server-side only; it uses your API key.
npm install brokerbot
import { BrokerBot } from "brokerbot"
const brokerbot = new BrokerBot({ apiKey: process.env.BROKERBOT_API_KEY! })
new BrokerBot() also accepts baseUrl (default https://api.brokerbot.ai) and a custom fetch. Failed requests throw a BrokerBotError with the HTTP status and the API’s error message.