Click https://huggingface.co/unsloth/Qwen3-4B-Instruct-2507-GGUF/resolve/main/Qwen3-4B-Instruct-2507-IQ4_XS.gguf?download=true Download the model Drop it in your Writingway/models folder

Running tests

This project includes a small test harness (smoke, unit, and UI tests) to help keep refactors safe.

Prerequisites

• Node.js (recommended >= 16)
• The UI tests use Playwright (included as a devDependency). If you run into missing browsers, run npx playwright install.
• If you want generation tests to exercise the real model, start the local AI server (the project expects it at http://localhost:8080). You can start the provided server with start.bat on Windows.

Available test scripts (run from the project root):

npm run smoke    # quick smoke: loads the page and checks key selectors
npm run unit     # unit tests (e.g., Generation.buildPrompt)
npm run ui       # runs the UI tests (headless)
npm test         # runs smoke, unit, then ui in sequence

Notes

• The repository vendors Alpine locally at src/vendor/alpine.min.js so tests can run offline.
• UI tests run headless and are written to prefer calling the app's handlers and to fall back to Dexie DB operations and stubbed generation when the AI server is not available.
• Do not remove the model download instructions above — the model binary is large and must be downloaded separately.

If you want me to wire up a small test-only API on the app (window.__test) to make tests even more deterministic, tell me and I will add it and update tests to use it.