Class AIPromptExtension

constructor

• new AIPromptExtension(contextUser: UserInfo): AIPromptExtension

  Parameters

  • contextUser: UserInfo

  Returns AIPromptExtension

AllowFormattingText

parse

• parse(parser: any, nodes: any, lexer: any): any

  Required, must implement this method to parse the template and return the results of a call to nodes.CallExtensionAsync()

  Parameters

  • parser: any
  • nodes: any
  • lexer: any

  Returns any

run

• run(context: any, params: any, body: any, callBack: NunjucksCallback): void

  Executes AI prompt processing using the configured language model.

  Parameter Mapping from CallExtensionAsync:

  • In parse(): new nodes.CallExtensionAsync(this, 'run', params, [body])
  • Results in: run(context, params, body, callBack)

  The key difference from TemplateEmbed is that we pass [body] as the 4th parameter to CallExtensionAsync, which shifts the parameter order to include both params and body as separate arguments. The body contains the prompt text between the opening and closing tags.

  Parameters

  • context: any

    Nunjucks template rendering context containing variables and data

  • params: any

    Parsed parameters from the opening tag (AIModel, AllowFormatting, etc.)

  • body: any

    Function that returns the rendered prompt content between {% AIPrompt %} and {% endAIPrompt %}

  • callBack: NunjucksCallback

    Async callback function to return the AI-generated response or errors

  Returns void

  Example

  {% AIPrompt AIModel="gpt-4", AllowFormatting=true %}
  Generate a personalized greeting for {{ user.name }} who works as {{ user.jobTitle }}.
  Make it professional but warm.
  {% endAIPrompt %}
  Copy