1

The API URL to send requests to ChatGPT. By default, use the official OpenAI endpoint (https://api.openai.com/v1), or a custom endpoint if hosting a compatible server.

2

The API key for authenticating with the ChatGPT service. Ensure the key is kept secure.

3

The AI model to use for generating responses (e.g., gpt-4 or gpt-3.5-turbo).

4

Communication API version.

5

The temperature setting for response generation, between 0 and 1. Lower values produce more deterministic responses, while higher values increase creativity.

6

The maximum number of tokens (words, fragments) the AI can generate in a response.

7

Reasoning effort.

8

The role assigned to the AI within the conversation context, such as assistant, system, or a custom description to guide responses.

9

The AI embed model to compute vectors

10

The text file that contains the knowledge of the agent

11

Enable long term memory using the embedding model

12

The list of defined tools.

13

Add, Remove or edit a tool.

14

Enable default tools (web search, web fetch, get time, get date, add / get memory).

1

Function name (Ex. SetCubePosition).

2

Detailed function description (Ex. Sets the 3D position and orientation of a cube in 3D world, world unit is meter, this also returns the new position).

3

Detailed function parameters (Ex. X Y Z position in floats, space separated and on a new line X Y Z orientation in degree).

4

The function return a value to the AI agent.

Request

Sends a request to the AI with history and context. Ideal for multi‑turn conversations.

Single request

One‑off request without full history. For quick, standalone questions.

Describe picture

Sends an image and text parameters to get a detailed visual description.

Set model

Selects the AI model (e.g., gpt‑3.5‑turbo or gpt‑4). Triggers an event on change.

Set creativity

Adjusts temperature to balance consistency and creativity.

Set max tokens

Sets the maximum number of tokens allocated to each response.

Set api key

Updates the Bearer authentication key.

Set api url

Modifies the base URL (sandbox, staging, custom hosting).

Set role

Defines the AI persona. Uses [Q]/[A] markers for conversation starters.

Reset history

Clears history, buffers, and cookies, then replays the initial role.

(Tool name) call

The agent invokes a custom function automatically detected from available tools.

Get models list

Retrieves the list of supported models (e.g., gpt‑4‑turbo, o1‑preview).

Set reasoning effort

Adjusts cognitive depth: low/fast, medium/balanced, or high/thorough.

Interrupt

Interrupts the current request and clears pending tool callbacks.

Enable memory

Enables semantic storage for intelligent context retrieval.

Disable memory

Disables semantic search. The AI uses only immediate history.

Reset memory

Clears and completely rebuilds the embeddings table. Useful after bulk data loading.

Requested input

Triggered before the response, provides the exact user prompt text.

Response stream

Streams each chunk of the SSE flow for character‑by‑character typing animations.

Reasoning stream

Triggered for reasoning models, delivers hidden thought traces.

Response end

Sent at the end of the main response, once all chunks are concatenated.

End reasoning

Triggered when the internal monologue of the reasoning model ends.

End of agent turn

Signals the complete end of an AI thinking cycle, with no pending tools.

Response

Main payload containing the final concatenated text in real‑time.

End stream

Signals the definitive closure of the SSE stream. Allows timer cleanup.

Api Key changed

Triggered when the Bearer key changes (without changing the URL).

AI model changed

Sent when switching models, providing the new model identifier.

Api Url changed

Triggered when the base URL changes (local dev or production cloud).

Response error

Transmits API errors (invalid JSON, no choices, incompatible tools).

Response end

Closes a specific response type. Often used with conditional end logic.

Get picture

Sends image data (JPEG paths or base64 blobs) directly into the 3D scene.