---
title: "Chat Or Question And Answer Using Retreival Agent"
method: POST
path: "/teams/{team_id}/bots/{bot_id}/chat-agent"
---

# Chat Or Question And Answer Using Retreival Agent

`POST /teams/{team_id}/bots/{bot_id}/chat-agent`

## Request

- Base URL: `https://api.docsbot.ai` (the document declares no server; this is the origin it was published from)
- URL: `https://api.docsbot.ai/teams/{team_id}/bots/{bot_id}/chat-agent`
- Auth: none declared

## Path parameters

- `team_id` string, required
- `bot_id` string, required

## Request body

- AgentAsk
  - `question` string, nullable
  - `history` unknown
  - `format` 'markdown' | 'text' | 'html', nullable
  - `full_source` boolean, nullable
  - `metadata` object, nullable
  - `testing` boolean, nullable
  - `autocut` union
    - boolean
    - integer
  - `context_items` integer
  - `audio` union: Audio user message as a data URL string or an object with base64 data and format.
    - string
    - AgentAudioInput
      - `data` string, required: Base64 encoded audio data
      - `format` string, required: Audio format extension supported by OpenAI, such as webm, mp3, or wav
  - `conversationId` string, required: Unique identifier for the conversation, pass a new UUID to start a new conversation
  - `channel` 'chat' | 'voice' | 'phone' | 'slack' | 'helpscout' | 'messenger' | 'whatsapp' | 'telegram', nullable: Internal conversation channel persisted on the conversation object.
  - `stream` boolean: Whether to stream the response with SSE or return it all at once
  - `human_escalation` boolean: Whether to enable the human escalation detection tool
  - `calendly` boolean: Whether to enable the Calendly booking tool
  - `calcom` boolean: Whether to enable the Cal.com booking tool
  - `tidycal` boolean: Whether to enable the TidyCal booking tool
  - `custom_buttons` boolean: Whether to enable custom button tools built from bot.tools.customButtons
  - `followup_rating` boolean: Whether to enable the follow-up rating tool
  - `document_retriever` boolean: Whether to enable the document retrieval tool for factual answers
  - `web_search` union: Web search mode: false disables, true enables cached mode (or bot default), and 'live' forces live search for this request.
    - boolean
    - 'live'
  - `openapi` boolean: Ignore this field, not yet implemented
  - `image_urls` string[], nullable: List of image URLs to include as context for this message
  - `model` string, nullable: Any valid OpenAI model string supported by DocsBot.
  - `default_language` string, nullable: The default language to use, if the language of the conversation is unclear. Use the locale code like 'en' or 'en-US'. Defaults to bot language.
  - `reasoning_effort` 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh' | 'max', nullable: Reasoning depth for the response. Requires authentication and the Personal plan or above to override default. 'none' (GPT-5.1) and 'minimal' (GPT-5) use the least reasoning, 'low' is light, 'medium' balanced, 'high' deeper, 'xhigh' deeper still (GPT-5.4/5.6/6), and 'max' deepest (GPT-5.6 Sol or GPT-6 Astra). Defaults: GPT-6 Luna → 'medium', GPT-6 Sol/Astra → 'low', GPT-5.6 Luna → 'low', other GPT-5.6 → 'none', GPT-5.4 nano → 'medium', GPT-5.4 mini → 'low', other versioned GPT-5.x → 'none', GPT-5 → 'minimal', other reasoning models → 'low', non-reasoning models → ignored.
  - `search_limit` integer, nullable: Maximum number of times the search_documentation tool can be called. Requires authentication to override default. Minimum: 1, Maximum: 4. Defaults to bot's searchDocumentationLimit or 2 (default is 2 for latency and token cost reasons).

## Response `200`

Successful Response

- unknown

## Other responses

- `422`: Validation Error

---

[API](https://skmtc.dev/docsbot/apis/docsbot-chat-api.md) · [All operations](https://skmtc.dev/docsbot/apis/docsbot-chat-api/llms.txt) · [OpenAPI document](https://skmtc.dev/docsbot/apis/docsbot-chat-api/revisions/336b782b49d5?raw)