.hud_eval.toml

# HUD Eval Configuration
# Command-line arguments override these settings

[eval]
# source = "hud-evals/SheetBench-50"
# agent = "claude"
# all = false  # Run all problems instead of just 1
# max_concurrent = 30
# max_steps = 10
# group_size = 1
# task_ids = ["checkout-smoke", "0"]  # slugs or 0-based indices
# verbose = true
# very_verbose = true
# auto_respond = true
# gateway = false  # Route LLM API calls through HUD Gateway

[agent]
# allowed_tools = ["computer", "playwright"]
# disallowed_tools = []

[claude]
# model = "claude-sonnet-4-5"
# max_tokens = 16384
# use_computer_beta = true

[openai]
# model = "gpt-4o"
# temperature = 0.7
# max_output_tokens = 4096

[gemini]
# model = "gemini-2.5-pro"
# temperature = 1.0
# top_p = 0.95

[gemini_cua]
# model = "gemini-2.5-computer-use-preview"
# temperature = 1.0
# top_p = 0.95
# excluded_predefined_functions = []

[openai_compatible]
# base_url = "http://localhost:8000/v1"
# model = "my-model"