Testing Strategy

Current State

The project has 7 unit test files covering core/ domain logic and markdown parsing. Tests use Mocha (suite/test) with Node.js assert. The test infrastructure uses @vscode/test-cli and @vscode/test-electron.

src/test/unit/
  core/
    domain/
      host.test.ts          — Host model: init, defaults, env export, dump formats
      user.test.ts          — UserCredential model: init, env export, dump formats
      finding.test.ts       — Finding parsing from markdown frontmatter
      graph.test.ts         — Relationship graph + Tarjan SCC
    env/
      collects.test.ts      — mergeCollects priority behavior
    markdown/
      fencedBlocks.test.ts  — Fenced code block extraction
      yamlBlocks.test.ts    — YAML block extraction by identity

---

Running Tests

pnpm run compile-tests    # Compile to out/ via tsc
pnpm run test:unit        # Run unit tests

In CI, use xvfb-run for headless VS Code test execution (Linux).

---

Testing Pyramid

Layer	What to Test	How	Target Coverage
core/	Domain models, parsing, graph algorithms	Pure unit tests, no mocks needed	90%+
platform/vscode/	Context caching, variable resolution, env collection	Mock workspaceState and vscode.workspace	60%+
features/	Command handlers, sync logic, MCP tool handlers	Mock Context + VS Code APIs	70%+
E2E	Activation, file sync, full command workflows	@vscode/test-electron with test workspace	Critical paths

---

What's Covered

The existing unit tests focus on the core/ layer:

• Host/User models: initialization with full/minimal/missing data, default values, environment variable export (exportEnvironmentCollects), dump format outputs (env, hosts file, yaml, table, impacket, nxc)
• Finding: parsing frontmatter fields (severity, tags, description) from markdown
• Graph: node/edge construction from Foam resources, Tarjan SCC computation, longest path calculation
• Collects: mergeCollects first-writer-wins behavior, priority ordering
• Markdown parsing: fenced block extraction (with/without language tags), YAML block extraction by identity keyword

---

Gaps

Area	Gap	Difficulty
platform/vscode/context.ts	Dirty-flag caching, Foam() lifecycle	Medium — needs workspaceState mock
platform/vscode/variables.ts	Variable resolution (${env:X}, ${config:X})	Easy — pure string processing
features/targets/sync	File watcher → state rebuild pipeline	Medium — needs filesystem + Context mock
features/mcp/httpServer.ts	MCP tool handlers return correct data	Medium — needs Context mock
features/ai/participant.ts	Prompt construction, command routing	Easy — mock vscode.lm
features/http/rawRequest.ts	HTTP request execution	Hard — needs HTTP mock

---

Writing Tests

Use Mocha suite/test with Node.js assert. Mirror the source directory structure under src/test/unit/:

Source: src/core/domain/host.ts
Test:   src/test/unit/core/domain/host.test.ts

Key patterns used in existing tests:

• suite("ClassName") groups related tests
• test("method() with scenario") names describe input conditions
• assert.equal / assert.deepEqual / assert.ok for assertions
• Direct instantiation — no DI framework, just new Host().init({...})
• No external test fixtures — test data is inline in each test file