# Evaluation for checking appropriate todo_write tool usage
run:
  - git clone --depth=1 --branch main https://github.com/tailcallhq/forgecode .
  - FORGE_OVERRIDE_PROVIDER=open_router FORGE_OVERRIDE_MODEL={{model}} FORGE_DEBUG_REQUESTS='{{dir}}/context.json' forgee -p '{{task}}'
parallelism: 3
timeout: 240
early_exit: true
validations:
  - name: "Uses todo_write tool for multi-step tasks"
    type: shell
    command: |
      # Check if todo_write was called at least once
      jq -e '[.messages[]?.tool_calls[]? | select(.function?.name == "todo_write")] | length > 0' {{dir}}/context.json

sources:
  - value:
      - model: "anthropic/claude-sonnet-4.5"
  - csv: todo_write_usage_tasks.csv