API Reference

Update agent settings.

post
https://app.customgpt.ai/api/v1/projects//settings

Updates an agent's configuration, including persona instructions, response format, citation style, and deployment settings.

Notes: Only include the fields you want to change — omitted fields retain their current values.

Recent Requests
Log in to see full request history
TimeStatusUser Agent
Retrieving recent requests…
LoadingLoading…
Path Params
integer
required

The unique identifier of the agent (formerly known as project).

Body Params

Update agent settings

file

This is the avatar that is shown in the agent response. You can make it a profile picture or your company logo.

file

This is the background image shown in the agent conversations widget. You can change it to a company logo or background image.

string
enum
Defaults to color

This is the background image shown in the agent conversations widget. You can change it to a company logo or background image.

Allowed:
string
Defaults to #FFFFFF

This is the background color shown in the agent conversations widget. You can change it to any color of your choice.

string
enum
Defaults to fresh

This is the color scheme of an agent.

Allowed:
string
enum
Defaults to soft

This is the radius styling of an agent.

Allowed:
string
enum
Defaults to inter

This is the font family of an agent.

Allowed:
string
length ≤ 255

This is the default prompt shown to the user. You can customize this for your company or client.

example_questions[]
array of strings

These are example questions shown to guide users. You can create customized questions to suit your company or client needs.

example_questions[]
string
enum
length ≤ 255

Controls which content sources the agent uses when generating responses. Possible values:

  • own_content: Answers only from indexed content. Returns a fallback message if the answer is not in the knowledge base.
  • default: Exists for backwards-compatibility. Same behavior as own_content.
  • openai_content: Uses the indexed knowledge base and supplements with general AI knowledge as needed.
Allowed:
string
enum
length ≤ 255

By default, the agent messages like 'Ask Me Anything' are in English. You can customize this to your preferred language. Please note: This setting does not control what language ChatGPT responds in. That is controlled by the user's question. So a user asking in Portuguese, will most likely get a response from ChatGPT in Portuguese.

string
length ≤ 9

Color of the agent in hex format

string
length ≤ 9

Color of the agent toolbar in hex format

string | null

Customize your agent behavior by adjusting the system parameter to control its personality traits and role.

string | null

This is the message shown to indicate where the response came from. You can customize this message based on your business or language.

string | null

This is the message shown for the Sources label. You can customize this message based on your business or language.

string | null

This is the message shown when the agent is thinking and waiting to answer. You can customize this message based on your tone, personality or language.

string | null

This is the message shown when the agent has encountered a problem or error. You can customize this message based on your tone, personality or language.

string | null

This is the message shown when the agent fails moderation due to inappropriate content. You can customize this message based on your tone, personality or language.

boolean | null
Defaults to true

Show animated loading indicator while waiting for a response from the agent

integer | null
enum
Defaults to 3

Controls how citations are displayed in agent responses. Possible values:

  • 0: Disabled - no citations shown
  • 1: End-of-response - citations appear inside a widget after the response
  • 2: Inline - citations appear as numbered references within the response text
  • 3: Both - inline references and an end-of-response widget
Allowed:
boolean | null
Defaults to true

Each agent response shows an thumbs up/down for the user to left own feedback.

boolean | null
Defaults to true

Controls the visibility of the action button that lets users quickly copy the agent's entire response.

string | null
enum
Defaults to user

Controls the initial citation visibility in the chat widget. Possible values:

  • user: Initially collapsed, but user's choice will be saved.
  • show: Citations in every message are opened by default.
  • hide: Citations in every message are collapsed by default.
Allowed:
string | null
enum
Defaults to default

Controls how images are displayed in citations. Default displays like regular citations, first_only inserts the image below the first paragraph where referenced.

Allowed:
boolean | null
Defaults to false

Limit the height of image citations.

boolean | null
Defaults to false

Use OpenGraph images from source pages when rendering image citations.

string | null
Defaults to I'm sorry, I don't know the answer

This is the message shown when the agent cannot answer. You can customize it to a message asking the user to contact customer support or leave their email / phone.

string | null

You can instruct ChatGPT to end every response with some text like asking "Please email us for further support" (Not recommended for most use cases)

string | null
Defaults to Try asking these questions...

This is the message shown to the user to encourage them to ask questions. You can customize this message based on your tone, personality or language.

string | null
Defaults to View more

This is the message shown to the user to view more starter questions. You can customize this message based on your tone, personality or language.

string | null
Defaults to View less

This is the message shown to the user to view less starter questions. You can customize this message based on your tone, personality or language.

boolean | null
Defaults to false

Controls what branding is shown at the bottom of the agent.

boolean | null
Defaults to false

Should we check messages from guests with Recaptcha when your agent is publicly available (i.e. shared or embedded).

string | null
enum

The AI model powering this agent. Available models are grouped by provider:

  • GPT (OpenAI): gpt-4-1, gpt-4-o, gpt-4-1-mini, gpt-4o-mini, and gpt-5.1, gpt-5.2 and gpt-5.4 variants
  • Claude (Anthropic): claude-4.6-opus, claude-4.6-sonnet, claude-4.5-haiku
  • Gemini (Google): gemini-3.1-pro, gemini-3-pro, gemini-2.5-flash, and variants

Model availability depends on your plan. See the CustomGPT dashboard for current availability, capability comparisons, and pricing.

boolean | null
Defaults to false

Enable selling of the agent for monetization

boolean | null
Defaults to false

Toggle conversation sharing.

boolean | null
Defaults to false

Toggle export sharing.

boolean | null
Defaults to true

Should the agent mention source names?

string | null
enum
Defaults to optimal-choice

The capability preset that determines model selection and reasoning depth. Possible values:

  • fastest-responses: Lightning-fast, sub-second responses.
  • optimal-choice: Best for most business use cases.
  • advanced-reasoning: Most relevant responses, recommended for large agents.
  • complex-tasks: Break down user queries for deeper understanding.
Allowed:
string | null

You can add URL for your Terms of Service here, and it will be automatically added to your agent's interface.

boolean
Defaults to false

Enable user avatar

file | null

User avatar

boolean
Defaults to true

Enable title avatar

boolean
Defaults to false

Enable spotlight avatar

file | null
Defaults to spotlight_avatar.png

Spotlight avatar

string | null
enum
Defaults to rectangle

Spotlight avatar shape

Allowed:
string | null
enum
Defaults to default

Spotlight avatar type

Allowed:
string | null
enum
Defaults to agent-left-user-right

Avatars orientation

Allowed:
string
enum
Defaults to left

Agent title & avatar alignment

Allowed:
string | null

Chatbot title

string | null
Defaults to #565656

Chatbot title color, HEX color code

boolean | null
Defaults to false

Should the citations be shown inline in the response?

integer | null
enum
Defaults to 0

Set how long conversations remain active. [0] 24 hour memory limit - conversations stay open but agents only process messages from the last 24 hours. [1] Unlimited - conversations never expire and agents see the full history. [2] 24 hours, then close - conversations lock 24 hours after it started and users can't send new messages.

Allowed:
string
enum
Defaults to year

Period after which we automatically delete all conversations.

Allowed:
integer
Defaults to 0

Custom period in days after which we automatically delete all conversations.

boolean | null
Defaults to false

Use context-aware starter question

boolean
Defaults to true

Enable agent knowledge base awareness

boolean | null
Defaults to true

Enable Markdown support in the agent responses. This allows for rich text formatting in the agent responses, such as bold, italics, lists, and links.

boolean | null
Defaults to false

When enabled, we take a thorough look at any numeric codes in user's prompt (e.g. ABCD1234) to find the best match in the knowledge base. Use this if your agent is struggling with, for example, product names from uploaded spreadsheet. This feature might make agent's responses slower.

boolean | null
Defaults to false

When enabled, advanced alphanumeric code extraction is applied at document indexing time via Pinecone metadata filtering for better numeric search results.

boolean
Defaults to true

When enabled, the agent will be aware of the logged-in user's name and can use it in conversations.

boolean | null

Enable or disable V3 agentic flow for this agent.

Responses

Updated 2 months ago

Language
Credentials
Bearer
JWT
LoadingLoading…
Response
Click Try It! to start a request and see the response here! Or choose an example:
application/json

Updated 2 months ago