CI

Run reagent-flow contract tests in CI without live LLM calls.

CI contract tests

reagent-flow tests should be deterministic. Do not make live LLM API calls in CI. Build contract tests with mocks, fixtures, or manually logged sessions, then fail the PR when a handoff or tool-output contract drifts.

Minimal command

uv run pytest tests/ -v

If your reagent-flow tests live in a specific folder:

uv run pytest tests/agent_contracts/ -v

GitHub Actions

name: Agent contract tests

on:
  pull_request:
    branches: [master]
  push:
    branches: [master]

jobs:
  contract-tests:
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v6

      - name: Install uv
        uses: astral-sh/setup-uv@v7

      - name: Set up Python
        uses: actions/setup-python@v6
        with:
          python-version: "3.12"

      - name: Install dependencies
        run: uv sync

      - name: Run contract tests
        run: uv run pytest tests/ -v

Recommended release checks

For a repository using reagent-flow itself, run:

uv run ruff check packages/ examples/
uv run ruff format --check packages/ examples/
uv run mypy packages/reagent-flow/src/reagent_flow/ --strict
uv run coverage run -m pytest packages/
uv run pytest examples/vendor_onboarding_showcase/test_showcase.py -v

Trace files in CI

Use a temporary trace_dir in tests unless you intentionally want trace files as artifacts. Do not upload .trace.json files from public CI unless they are sanitized.

For sensitive keys, use redact_fields:

with reagent_flow.session(
    "security-review",
    trace_dir=str(tmp_path),
    redact_fields={"api_key", "customer_email"},
) as s:
    ...

Redaction only affects saved trace files. Assertions still validate the original in-memory values.