Skip to main content
FilDOS is tested at three levels, all runnable locally and all enforced in CI.

​
Unit + integration (Vitest)

Tests live beside the code as *.test.ts, split per feature, reusing the Vite aliases via vitest.config.ts:
  • Renderer tests run under jsdom.
  • Main-process tests run under node.
The filesystem service tests (src/main/fs/service.*.test.ts, one file per operation — create, read, rename, copy, move, …) are true integration tests: they exercise real files in an os.tmpdir() sandbox provisioned by src/main/fs/fixtures.ts. The AI layer is tested the same way: dependencies are injected into Indexer, IndexWatcher, and GraphBuilder, so tests drive the whole pipeline with a fake provider and a temp directory — no Electron required. Chat tools are tested against a temp dir in tools.test.ts. 
npm test            # vitest run — unit + integration
npm run test:watch  # vitest in watch mode

​
End-to-end (Playwright)

e2e/*.spec.ts launch the built Electron app via _electron (playwright.config.ts) and assert the shell boots — so test:e2e builds first. 
npm run test:e2e    # build, then Playwright smoke test on the packaged app
On a headless Linux box, run the e2e suite under xvfb-run.

​
Lint + hooks

ESLint uses a flat config (eslint.config.mjs) that splits environments along the process boundary — node vs. browser, plus React Hooks rules. A git pre-commit hook runs lint-staged (eslint --fix on staged files); the prepare script installs it on npm install. 
npm run lint        # eslint (flat config)
npm run typecheck   # tsc for both node (main/preload) and web (renderer) projects

​
The full pre-PR loop

Before opening a pull request, run: 
npm run lint
npm run typecheck
npm test
npm run build

​
Continuous integration

.github/workflows/ci.yml runs lint, typecheck + unit/integration, and e2e on Ubuntu, macOS, and Windows, on Node 22 (required for node:sqlite).