promptfoo--promptfoo/examples/eval-custom-grading-prompt/promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
description: Custom LLM-based grading with evaluation prompts

prompts:
  - file://prompts.txt
providers:
  - openai:gpt-4.1-mini
defaultTest:
  options:
    # ---
    # Custom grading prompt:
    # ---
    rubricPrompt:
      - role: system
        content: >-
          Grade the output by the following specifications, keeping track of the points scored:

          Did the output mention {{x}}? +1 point
          Did the output describe {{y}}? + 1 point
          Did the output ask to clarify {{z}}? +1 point

          Calculate the score but always pass the test. Output your response in the following JSON format:
          {pass: true, score: number, reason: string}
      - role: user
        content: 'Output: {{ output }}'

#    ---
#    You can also provide an OpenAI prompt directly as pure JSON:
#    ---
#
#    rubricPrompt: >-
#      [
#        {
#          "role": "system",
#          "content": "Grade the output by the following specifications, keeping track of the points scored:\n Did the output mention {{x}}? +1 point\n Did the output describe {{y}}? + 1 point\n Did the output ask to clarify {{z}}? +1 point\n\n Output your response in the following JSON format:\n {pass: bool, score: number, reason: string}"
#        },
#        {
#          "role": "user",
#          "content": "Output: {{ output }}"
#        }
#      ]
#

tests:
  - vars:
      topic: the economy
      x: the Federal Reserve
      y: macroeconomics
      z: the specific question the user wants to ask
    assert:
      - type: llm-rubric