Configure Pytest
Adapter status: ⛔ STUB. The Pytest adapter is documented but not implemented. pytest is a valid value in qa-stack.yaml and lippia-neo-bootstrap-stack detects it, but the skills that need the adapter stop as soon as they start:
Adapter pytest en estado STUB.
Ver <skill_root>/adapters/pytest/STUB.md para contribuir.
Value in qa-stack.yaml |
test_framework: pytest |
| Language | python |
| Intended app type | Mainly api. Pytest has no browser of its own, and the pytest-playwright case isn't covered |
| Default tests folder | tests/pytest/ |
1. What works today and what doesn't
| Skill | Does it work with test_framework: pytest? |
|---|---|
lippia-neo-bootstrap-stack |
✅ Detects Pytest (pyproject.toml, setup.cfg, or requirements.txt with pytest) and generates qa-stack.yaml |
lippia-neo-read-user-story, lippia-neo-create-test-plan, and the rest of the skills that don't touch test code |
✅ They don't depend on the framework |
lippia-neo-generate-test-suite |
⛔ Stops: it can't generate tests |
lippia-neo-run-and-heal |
⛔ Stops: it can't run or fix tests |
| Playbooks that generate or run tests (Feature End-to-End, Bug Fix Cycle, Smoke Test, Hotfix…) | ⛔ They stop at the first step that uses those skills |
In the meantime: if what you need is to test an API, the Playwright adapter generates and runs API tests (
app_type: api) without a browser. See Configure Playwright.
2. Get the project ready
If your project is Python and you want to have it ready for when the adapter is implemented, you can generate qa-stack.yaml right now:
stack:
test_framework: pytest
language: python
app_type: api
auth: api_token
pipeline_provider: gitlab
environment:
base_url: https://api.techstore-staging.com
app_version: sprint-24
os: Ubuntu 22.04
api_contract:
spec_path: openapi.yaml
spec_url: null
paths:
user_stories_storage:
type: local
cloud_provider_mcp: null
local_path: user-stories/
test_plans_dir: test-plans/
tests_dir: tests/pytest/
reports_dir: test-results/
env_file: .env.tests
commands:
run_tests: pytest --junitxml=test-results/results.xml
install_deps: pip install -r requirements.txt
And have installed:
| Tool | Intended version |
|---|---|
| Python | 3.9 or higher |
pytest |
— |
pytest-json-report |
To generate the JSON that parse-results.sh will read |
httpx (or requests) |
For the API tests |
pip install pytest pytest-json-report httpx
3. How it will work (intended design)
According to the contract documented in each skill's STUB.md, once the adapter is implemented:
| Aspect | Intended design |
|---|---|
| Generated files | tests/<US-ID>/test_tc_NN_<slug>.py, with functions in snake_case |
| Authentication | conftest.py fixtures with httpx.Client: one with the manager token and another with the viewer token |
| Credentials | Through environment variables (API_BASE_URL, API_TOKEN_MANAGER, API_TOKEN_VIEWER, QA_API_TOKEN), never literals |
| Cleanup | An autouse=True fixture that deletes the resources created by the tests |
| Assertions | Standard Python assert |
| Run command | pytest <tests_dir> --junitxml=test-results/results.xml --json-report --json-report-file=test-results/results.json -v |
| Results | parse-results.sh converts test-results/results.json to NEO's common format |
Example of the intended template for an API test:
import os
import pytest
import httpx
BASE_URL = os.environ["API_BASE_URL"]
TOKEN_MANAGER = os.environ["API_TOKEN_MANAGER"]
@pytest.fixture
def manager_client():
headers = {"Authorization": f"Bearer {TOKEN_MANAGER}", "Content-Type": "application/json"}
with httpx.Client(base_url=BASE_URL, headers=headers) as client:
yield client
def test_tc_01_crear_orden_happy_path(manager_client):
response = manager_client.post("/orders/v1", json={"product_id": 42, "quantity": 1})
assert response.status_code == 201
4. What's missing to complete it
For anyone who wants to contribute to the skills repository, these need to be created:
| Skill | Files to create in adapters/pytest/ |
|---|---|
lippia-neo-generate-test-suite |
README.md (conventions), test-template.py (API test template), and conftest-template.py (shared fixtures) |
lippia-neo-run-and-heal |
run.sh (runs pytest), parse-results.sh (converts the pytest-json-report JSON to the common format), and README.md |
Once the README.md exists in the adapter folder, the skills stop treating it as a STUB and use it like any other. The full details are in skills/lippia-neo-generate-test-suite/adapters/pytest/STUB.md and skills/lippia-neo-run-and-heal/adapters/pytest/STUB.md in the skills repository.
5. Common problems (and what to do)
| What you see | Why it happens | What to do |
|---|---|---|
Adapter pytest en estado STUB |
The adapter isn't implemented yet | Use Playwright for API tests in the meantime, or contribute the adapter (section 4) |
| A playbook stops when generating or running tests | All playbooks use lippia-neo-generate-test-suite or lippia-neo-run-and-heal |
Same as the previous case |