test.yml

name: tests

on:
  push:
    branches: [main]
  pull_request:

jobs:
  test:
    runs-on: ${{ matrix.os }}
    strategy:
      fail-fast: false
      matrix:
        # Both OSes on purpose: the cp1252 console bug and the Windows
        # drive-letter mount parsing are only reachable on Windows, and the
        # Docker tests only run on the Linux runners.
        os: [ubuntu-latest, windows-latest]
        python: ["3.11", "3.13"]

    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-python@v5
        with:
          python-version: ${{ matrix.python }}

      - name: Install
        run: |
          python -m pip install --upgrade pip
          pip install -e ".[dev]"

      - name: Lint
        run: ruff check agent adapters tests

      - name: Test
        # Docker tests skip themselves when no daemon is reachable, so this is
        # the same command everywhere. Ubuntu runners have one; Windows does not.
        run: python -m pytest -q

      - name: Terminal-Bench adapter
        # Needs Python >= 3.12, so it only runs on the newer matrix entry.
        if: matrix.python == '3.13'
        run: |
          pip install terminal-bench
          python -m pytest tests/test_adapter.py -q