BXL's Copilot Skill to test 5g functions

function-test

---
name: function-test

description: Tests a custom Excel function by randomizing input arguments and verifying its spilled output produces identical results to a reference formula.
--- 

metadata:
  version: 1.0.0
  tags:
    - excel
    - lambda
    - afe
    - excel labs
    - named functions
    - documentation
    - 5g modeling

triggers:
  - test
  - test function
  - function test

rules:
  - Work step-by-step.
  - After completing each step, pause and show results including the affected sheet/range and key values before continuing.
  - When randomizing a timeline, randomize the timeline's start, periods, and interval arguments and display these arguments in the results table, not the resulting dates in the timeline. 
  - When running the tests, if any fail for any reason stop the tests, record all tests in the table including the failed test.



workflow:

  - step 1:
    display "SKILL: function-test VERSION: BXL 2026-09-02" then continue.

  - step 2: 
    name: gather_information
    If a named range "Language" does not exist:
    prompt: "Which language would you prefer to use?" in the regional setting's likely language.
      - Display options as a numbered list:
        1 English
        2 Español
        3 Français
        4 Deutsch
        5 Italiano
        6 Português
        7 中文
        8 日本語
        9 한국어
        10 Other — type the language you want to use.
      - Switch to the language and use it for the remaining steps and worksheet entries.

  - step: 3
    name: get_function_cell
    action: prompt_user
    prompt: "Select the cell containing the function to test. When selected type C and press ENTER to continue"
    output: function_cell

  - step: 4
    name: get_comparison_cell
    action: prompt_user
    prompt: "Select the cell containing the formula to compare against. When selected type C and press ENTER to continue"
    output: comparison_cell

  - step: 5
    name: parse_function
    action: extract_formula_info
    input: function_cell
    output:
      function_name: function_name
      argument_names: argument_names
      argument_cells: argument_cells

  - step: 6
    name: get_test_parameters
    action: prompt_user
    prompt: "How many tests should I perform (enter a positive whole number)?"
    output: test_count

  - step: 7
    name: run_tests
    action: execute_test_loop
    inputs:
      function_cell: function_cell
      comparison_cell: comparison_cell
      argument_cells: argument_cells
      test_count: test_count

    process:
      - for_each_iteration:
          - randomize_arguments:
              method: "generate random values based on inferred cell data types"

          - wait_for_calculation

          - get_spilled_range:
              source: function_cell
              output: function_results

          - get_spilled_range:
              source: comparison_cell
              output: comparison_results

          - compare_results:
              method: element_wise
              tolerance:
                value: 0.0001
                type: absolute
              output: test_results

  - step: 8
    name: report_results
    action: generate_report
    input: test_results
    format:
    type: worksheet_table
    destination: worksheet
    columns:
    - Test
    - Pass
    - {argument_names}
    description: >
    Create a table directly in the worksheet where each row represents one test iteration.
    The table must include:
    - “Test” column containing the test iteration number
    - “Pass” column indicating whether the test passed or failed
    - one column for each randomized function argument, using the argument names as headers

    Each row should populate the argument columns with the exact values used during that iteration, so the full input set for each test is visible alongside its result.