Configure a skill prompt

  • Release version: Zurich
  • Updated July 31, 2025
  • 1 minute to read

    • Configure your skill prompt to set the model that is used and the randomness and creativity of the response.

    Before you begin

    Role required: sn_skill_builder.admin

    Procedure

    1. Navigate to All > Now Assist Skill Kit > Home.
    2. Select the skill that you want to configure.
    3. Select Configurations.
      Configuration panel for the Now Assist Skill Kit prompt.
    4. On the form, fill in the fields.
      Table 1. Configurations form
      Field Description
      Model The model is the large language model (LLM) that you want to use for the prompt.
      Temperature The temperature determines the randomness and creativity of the output. A higher value increases the randomness. The value must be between 0-1.
      Thinking mode Thinking mode controls how much reasoning effort a large language model applies when it generates a response. Higher levels improve output quality but increase latency.

      You can select thinking mode when you use llm_generic_small_v2 or llm_generic_large_v2.
      Maximum response tokens The maximum number of tokens the model can return. If you’re using Now LLM Service, the maximum is 1000.
      Maximum request tokens The maximum number of tokens allowed in a request.
      Structured output
      Structured outputs return prompt responses in a consistent JSON format.
      Note:
      Only Google Gemini and Azure OpenAI support structured output.
    5. Add Usage conditions to determine when to use the prompt.

    What to do next

    After you configure your skill settings, you can test your skill. To learn more about testing skills, see Test a prompt.

    To learn more about configuring the skill, including models and tokens, see Now Assist Skill Kit FAQs on the ServiceNow Community.