Testing

The wiremd test suite runs through Turborepo across the monorepo. The bulk of the tests live in packages/core/; smaller suites cover the web editor (apps/editor/) and the Figma plugin (extensions/figma/). All tests use Vitest.

Running tests

All commands run from the repo root.

# All tests across the monorepo (cached by turbo, runs in parallel where possible)
pnpm turbo run test

# Only the core library
pnpm --filter @eclectic-ai/wiremd run test

# Watch mode (core)
pnpm --filter @eclectic-ai/wiremd run test:watch

# Coverage (core)
pnpm --filter @eclectic-ai/wiremd run test:coverage

# Single test file (run from packages/core/)
pnpm --filter @eclectic-ai/wiremd run test -- tests/parser.test.ts

# Single test by name
pnpm --filter @eclectic-ai/wiremd run test -- tests/parser.test.ts -t "Button"

# Editor tests only
pnpm --filter wiremd-editor run test

# Figma plugin tests only
pnpm --filter wiremd-figma-plugin run test

pnpm turbo run test depends on build (see turbo.json), so the core library is built first, then test runs are fanned out to every workspace that defines a test script.

Test layout

packages/core/tests/

The main suite. Notable files:

• parser.test.ts — Markdown + wiremd-syntax parsing
• renderer.test.ts — HTML rendering
• react-renderer.test.ts — React/JSX renderer
• tailwind-renderer.test.ts — Tailwind-classed HTML
• integration.test.ts — end-to-end parse → render
• cli.test.ts, cli-unit.test.ts — CLI behaviour, watch mode, dev server
• cli-bundle.test.ts — the standalone CLI bundle
• server.test.ts — live-reload dev server, WebSocket injection, viewport switcher
• error-handling.test.ts, validation.test.ts — error and validation paths
• api-examples.test.ts — the public API examples in the docs
• package-shape.test.ts — published package shape (entry points, exports, files)

apps/editor/tests/

Browser editor: storage adapters, file-system access shim, render-markup glue.

extensions/figma/tests/

Figma plugin: AST → Figma frame conversion, theme styling, fixture-based regression tests. Mocks the Figma global API in tests/setup.ts.

Writing tests

Where new tests go

Change	Test home
Parser / renderer / CLI / library API	packages/core/tests/
Editor UI behaviour	apps/editor/tests/
Figma converter	extensions/figma/tests/
Cross-package integration	packages/core/tests/integration.test.ts

Style

// packages/core/tests/feature.test.ts
import { describe, it, expect } from 'vitest';
import { parse } from '../src/parser';
import { renderToHTML } from '../src/renderer';

describe('Feature name', () => {
  it('renders the basic case', () => {
    const html = renderToHTML(parse('## Heading'), { style: 'sketch' });
    expect(html).toContain('<h2');
  });

  it('handles the edge case', () => {
    // Arrange / Act / Assert
  });

  it('throws for invalid input', () => {
    expect(() => parse('')).toThrow();
  });
});

Use descriptive names (it('renders viewport switcher buttons', …)), not opaque ones (it('test viewport', …)). Async tests use async/await.

Continuous integration

The ci.yml workflow runs on every push to main and every pull request. Jobs:

Job	Matrix	Runs
test	Node 20 + 22, Linux + macOS + Windows	pnpm turbo run build, typecheck, lint, test
coverage	Node 20 / Linux	pnpm --filter @eclectic-ai/wiremd run test:coverage, uploads to Codecov
lint-package	Node 20 / Linux	Builds the core, runs npm pack to verify the published package shape
test-editor	Node 20 / Linux	pnpm turbo run test --filter=wiremd-editor
test-figma-plugin	Node 20 / Linux	pnpm turbo run test --filter=wiremd-figma-plugin

A green PR means: every workspace's tests pass, the core compiles + lints cleanly on three OSes, and the published shape is intact.

Local pre-push hook

pnpm install configures Git to use the repo's .githooks/ directory. The pre-push hook runs:

pnpm --filter @eclectic-ai/wiremd build
pnpm --filter @eclectic-ai/wiremd test:coverage

This catches fixture snapshot drift and core build failures before a branch is pushed. Run pnpm run prepare if hooks need to be re-enabled after cloning or changing Git config.

Coverage

pnpm --filter @eclectic-ai/wiremd run test:coverage

Coverage targets (current goals, not enforced):

• Overall: 80%+
• Critical paths (parser, html-renderer, CLI argv): 90%+
• Error handling: 100%

Reports land in packages/core/coverage/ and are uploaded to Codecov from the coverage CI job.

Common patterns

Temporary files

import { mkdtempSync, rmSync, writeFileSync } from 'node:fs';
import { tmpdir } from 'node:os';
import { join } from 'node:path';

let dir: string;

beforeEach(() => {
  dir = mkdtempSync(join(tmpdir(), 'wiremd-'));
  writeFileSync(join(dir, 'test.md'), '# Test');
});

afterEach(() => {
  rmSync(dir, { recursive: true, force: true });
});

Always use os.tmpdir() instead of /tmp so tests work on Windows runners (this was a real CI fix — see commit f0e4164).

Testing the CLI

import { execSync } from 'node:child_process';

it('generates output', () => {
  const out = execSync(
    'node packages/core/dist/cli/index.js input.md',
    { encoding: 'utf-8' },
  );
  expect(out).toContain('Generated');
});

Mocking the dev server

import { vi } from 'vitest';

vi.mock('http', () => ({ createServer: vi.fn() }));

Debugging

Run a single test by name

pnpm --filter @eclectic-ai/wiremd run test -- -t "should render sketch style"

Verbose reporter

pnpm --filter @eclectic-ai/wiremd run test -- --reporter=verbose

Debug under VS Code

.vscode/launch.json:

{
  "type": "node",
  "request": "launch",
  "name": "Debug wiremd tests",
  "runtimeExecutable": "pnpm",
  "runtimeArgs": ["--filter", "@eclectic-ai/wiremd", "run", "test", "--", "--no-coverage"],
  "console": "integratedTerminal"
}

When tests are slow

Vitest reports per-test timings. Tests over a second or two should either be optimized or moved to integration.test.ts so they can be filtered out with --exclude.

pnpm --filter @eclectic-ai/wiremd run test -- --reporter=verbose