Chat

Plugin ID: ai_integration_eca_execute_chat

Run text through the AI chat model.

Fields

Model — config key: model

The applicable model for this operation type.

Type: select · Required: yes · Default: empty string

Allowed values depend on the configuration of the target site.

Value format: The machine name of a model configured on the target site.

Token result — config key: token_result

The response from AI will be stored into the token result field to be used in future steps. Please provide the token name only, without brackets.

Type: textfield · Required: no · Default: empty string

Specific configuration for the model — config key: config

Some models require specific configuration settings, like temperature, voice or response_format.
Eg.

temperature: 0.5

Type: textarea · Required: no · Default: empty string

System prompt — config key: system_prompt

An optional instruction that sets the behavior of the model, such as "You are a helpful assistant". It is sent as the system prompt of the request, next to the prompt below, and supports the same tokens.

Type: textarea · Required: no · Default: empty string

Prompt — config key: prompt

Enter your text here. When submitted, AI will generate a response from its Chats endpoint. Based on the complexity of your text, AI traffic, and other factors, a response can sometimes take up to 10-15 seconds to complete. Please allow the operation to finish. Be cautious not to exceed the requests per minute quota (20/Minute by default), or you may be temporarily blocked.

Type: textarea · Required: yes · Default: Enter your prompt for AI here.

Schema — config key: schema

Provide an optional JSON schema to request structured output. When set, the model is asked to return JSON matching the schema, and that JSON is stored as text in the token specified in "Token result". Follow OpenAI's structured-output JSON Schema format: JSON Schema

Type: ai_json_schema · Required: no · Default: empty string

Files — config key: files

Optional file(s) to attach to the prompt. Accepts a single token holding a list of files (e.g. [my_files] or [node:field_images]), a single token holding one file, or a value typed in directly. Every value is read as configured in "Input source" below.
This field supports tokens.

Type: textfield · Required: no · Default: empty string

Input source — config key: files_input_type

How each value in "Files" should be interpreted. All attached files are read the same way.

Type: select · Required: yes · Default: path

Allowed values: fid, path, binary