For Developers

Testing in Sila

We have automated tests in two packages: core and the UI workbench.

packages/core/tests is for testing core (without the UI or any client-server logic).

packages/workbench/tests is for testing svelte components. The closest we get to e2e testing (via the UI workbench).

Perhaps we need to setup a proper testing for the packages/desktop (Electron app) but didn't do it yet.

Testing frameworks

The core tests use Vitest and the workbench uses Playwright.

Run tests

From the repository root:

bash

    # Install deps
npm install

# Run all top-level tests (core + workbench)
npm test

# Run only core tests
npm -w packages/core run test

# Run a single core test file (example)
npm -w packages/core run test -- --run src/ai/ai-image.test.ts

# Optional: use a specific config (e.g., node environment)
npx vitest --config packages/core/tests/vitest.config.ts

Playwright (workbench)

bash

    # Run workbench Playwright tests
npm -w packages/workbench run test

# Open Playwright UI runner
npm -w packages/workbench run test:ui

Playwright browser install (required once per machine/CI image)

Playwright tests need a browser binary (Chromium) downloaded locally.

bash

    # Local dev (Chromium only)
npx playwright install chromium

# CI / Linux runners (installs Chromium + OS deps)
npx playwright install --with-deps chromium

AI provider keys (optional)

Some tests use AI providers. Add API keys to enable them; tests auto-skip without keys.

Create .env in repo root
Supported keys (set whichever you use):
- OPENAI_API_KEY=...
- OPENROUTER_API_KEY=...

Test assets

Small, deterministic fixtures live in:

    packages/core/tests/assets/images/
packages/core/tests/assets/to-send/

Troubleshooting

If postinstall scripts are blocked:

bash

    npm install --ignore-scripts
npm -w packages/core run test

If Playwright fails with "Executable doesn't exist ... run npx playwright install":
bash
```
    npx playwright install chromium
  
```
AI tests without keys are expected to skip.