🧪 Testing & Quality Assurance

Reliable LLM systems don’t happen by accident—they’re built, broken, and verified. Locentra OS ships with a modular test suite that covers everything from real-time inference to memory scoring and agent workflows.

Tests are written with pytest and structured for clarity, speed, and layered validation.

---

✅ What’s Currently Covered

Subsystem	Coverage Focus
Core Config	Registry, settings, flags
Inference Engine	Prompt → Model → Output
Memory Layer	Vector embedding + recall
Agents	Retraining + feedback loop
API Routes	CRUD, responses, edge-cases
CLI Tools	Live queries, fine-tuning

Coverage improves with every release — current targets defined for v1.2 below.

---

🚀 Running the Tests

From the project root, run:

cd backend
pytest tests/

Need more detail?

pytest -v tests/

You’ll get per-test output including tracebacks for failed assertions.

---

🧰 Utility Structure & Fixtures

All test scaffolding lives in:

backend/tests/conftest.py

Includes:

• Dummy prompt builders

• Temporary memory inserts

• Mocked response objects

• Config overrides (e.g. fake tokens or models)

Use this file to globally patch components or simulate failure modes (e.g. invalid registry keys or missing memory).

---

🧪 Test Snippet (Example)

Here’s how a basic inference test looks in tests/test_infer.py:

from backend.core.registry import registry

def test_inference_response():
    prompt = "Explain zk-rollups"
    model = registry.get("model")
    output = model(prompt)
    assert "rollup" in output.lower()

No mocking required—Locentra’s test suite uses the same model and runtime paths as production by default.

---

⚙️ Writing Custom Tests

Want to extend coverage? It’s simple:

1. Create or extend a file in tests/

2. Import what you need from:

   • core/

   • models/

   • agents/

   • api/

3. Use standard assert syntax to validate outputs

4. Run pytest before committing

---

📈 Coverage Goals (v1.2)

Module	Target Coverage
Core Config / Registry	100%
Inference + Adapters	90%+
API Routes	90%
CLI Tools	60%+
Agent Logic	75%
Memory Scoring / Vectors	85%

Automated coverage reports (pytest-cov) coming in future milestone.

---

🤖 CI Integration (Planned)

To run tests automatically on every push:

# .github/workflows/tests.yml
name: Locentra Tests

on: [push]

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v2
      - name: Setup Python
        uses: actions/setup-python@v2
        with:
          python-version: '3.10'
      - name: Install deps
        run: pip install -r backend/requirements.txt
      - name: Run Pytest
        run: pytest backend/tests/

Add a .test.env to configure test environment settings, and mock any outbound model loading calls if needed.

---

🧪 Test Philosophy

“If it breaks, it should be visible immediately.”

Locentra favors:

• Small over large tests

• Live-run validation over mock-only

• Semantic memory lifecycle testing

• Trigger-based agent simulations (coming soon)