TypeScript-based integration tests for the awf (Agentic Workflow Firewall) CLI.

This directory contains comprehensive integration tests that verify firewall behavior across multiple scenarios. Currently includes 17 integration test files covering:

• Basic Firewall Functionality (basic-firewall.test.ts) - Domain whitelisting, subdomain matching, exit code propagation
• Exit Code Propagation (exit-code-propagation.test.ts) - Comprehensive exit code handling tests
• Container Working Directory (container-workdir.test.ts) - Container workdir configuration

• Blocked Domains (blocked-domains.test.ts) - Domain blocking and precedence
• Wildcard Patterns (wildcard-patterns.test.ts) - Wildcard pattern matching (*.domain.com)

• Network Security (network-security.test.ts) - Capability restrictions, bypass prevention, SSRF protection
• Robustness Tests (robustness.test.ts) - Edge cases, protocol handling, security corners

• DNS Servers (dns-servers.test.ts) - DNS server configuration and resolution
• Environment Variables (environment-variables.test.ts) - Environment variable passing
• Volume Mounts (volume-mounts.test.ts) - Volume mount configuration

• Protocol Support (protocol-support.test.ts) - HTTP/HTTPS, HTTP/2, IPv4/IPv6
• Git Operations (git-operations.test.ts) - Git clone, fetch, ls-remote

• Error Handling (error-handling.test.ts) - Network errors, command failures, recovery
• Log Commands (log-commands.test.ts) - Log parsing and analysis

• Claude Code (claude-code.test.ts) - Claude Code CLI integration
• No Docker (no-docker.test.ts) - Docker-in-Docker removal verification
• Docker Warning (docker-warning.test.ts) - Docker command warning messages

The firewall is tested via agentic workflow smoke tests that run through the actual firewall:

• Smoke Claude (.github/workflows/smoke-claude.md) - Claude engine validation
• Smoke Copilot (.github/workflows/smoke-copilot.md) - Copilot engine validation

These smoke tests use the locally built firewall and validate:

• GitHub MCP functionality
• Playwright browser automation
• File I/O operations
• Bash command execution

tests/
├── integration/              # Integration test suites (17 files)
│   ├── basic-firewall.test.ts
│   ├── blocked-domains.test.ts
│   ├── claude-code.test.ts
│   ├── container-workdir.test.ts
│   ├── dns-servers.test.ts
│   ├── docker-warning.test.ts
│   ├── environment-variables.test.ts
│   ├── error-handling.test.ts
│   ├── exit-code-propagation.test.ts
│   ├── git-operations.test.ts
│   ├── log-commands.test.ts
│   ├── network-security.test.ts
│   ├── no-docker.test.ts
│   ├── protocol-support.test.ts
│   ├── robustness.test.ts
│   ├── volume-mounts.test.ts
│   └── wildcard-patterns.test.ts
├── fixtures/                 # Reusable test utilities
│   ├── cleanup.ts            # Docker resource cleanup
│   ├── awf-runner.ts         # Execute awf commands
│   ├── docker-helper.ts      # Docker operations
│   ├── log-parser.ts         # Parse Squid/iptables logs
│   └── assertions.ts         # Custom Jest matchers
├── setup/
│   ├── jest.integration.config.js  # Jest configuration
│   └── jest.setup.ts               # Test setup
└── README.md                 # This file

1. Build the project:

   npm run build

2. Install dependencies:

   npm install

3. Ensure Docker is running:

   docker ps

4. Ensure sudo access: Tests require sudo for iptables manipulation.

# Unit tests + Integration tests
npm run test:all

npm test:unit

npm run test:integration

# Run volume mount tests
npm run test:integration -- volume-mounts

# Run container workdir tests
npm run test:integration -- container-workdir

npm run test:integration -- -t "Test 1: Basic volume mount"

Helper for executing awf commands:

import { createRunner } from '../fixtures/awf-runner';

const runner = createRunner();

// Run with sudo (required for iptables)
const result = await runner.runWithSudo('curl https://github.com', {
  allowDomains: ['github.com'],
  logLevel: 'debug',
  keepContainers: false,
});

// Check result
expect(result).toSucceed();
expect(result.exitCode).toBe(0);

Helper for Docker operations:

import { createDockerHelper } from '../fixtures/docker-helper';

const docker = createDockerHelper();

// Pull image
await docker.pullImage('curlimages/curl:latest');

// Run container
await docker.run({
  image: 'curlimages/curl:latest',
  command: ['curl', 'https://github.com'],
  rm: true,
});

// Inspect container
const info = await docker.inspect('awf-squid');
const isRunning = await docker.isRunning('awf-squid');

Parser for Squid and iptables logs:

import { createLogParser } from '../fixtures/log-parser';

const parser = createLogParser();

// Read and parse Squid logs
const entries = await parser.readSquidLog(workDir);

// Filter by decision
const allowed = parser.filterByDecision(entries, 'allowed');
const blocked = parser.filterByDecision(entries, 'blocked');

// Check if domain was allowed/blocked
const wasAllowed = parser.wasAllowed(entries, 'github.com');
const wasBlocked = parser.wasBlocked(entries, 'example.com');

Cleanup utility for Docker resources:

import { cleanup } from '../fixtures/cleanup';

// Run full cleanup
await cleanup(true); // true = verbose output

Custom Jest matchers for firewall assertions:

import { setupCustomMatchers } from '../fixtures/assertions';

setupCustomMatchers();

// Use custom matchers
expect(result).toSucceed();
expect(result).toFail();
expect(result).toExitWithCode(42);
expect(result).toAllowDomain('github.com');
expect(result).toBlockDomain('example.com');
expect(result).toTimeout();

Jest configuration for integration tests is in tests/setup/jest.integration.config.js:

module.exports = {
  preset: 'ts-jest',
  testEnvironment: 'node',
  roots: ['<rootDir>/../integration'],
  testMatch: ['**/*.test.ts'],
  testTimeout: 120000, // 2 minutes per test
  maxWorkers: 1, // Run tests serially to avoid Docker conflicts
};

Tests are designed to run in GitHub Actions. See .github/workflows/test-coverage.yml for the workflow configuration.

Key considerations:

• Tests run with sudo -E to preserve environment variables
• Docker images are pre-pulled to avoid timeouts
• Cleanup runs before and after tests to prevent resource leaks
• Artifacts (logs, reports) are collected on failure

The project uses TypeScript-based integration tests that run in CI via .github/workflows/test-coverage.yml:

Integration test files (17 total):

Smoke test workflows:

• .github/workflows/smoke-claude.md - Claude engine validation (uses locally built firewall)
• .github/workflows/smoke-codex.md - Codex engine validation (uses locally built firewall)
• .github/workflows/smoke-copilot.md - Copilot engine validation (uses locally built firewall)

CI workflow:

• All tests run with sudo -E for iptables manipulation
• Tests run serially to avoid Docker resource conflicts
• Automatic cleanup before and after test runs
• Test logs uploaded as artifacts on failure

Ensure you're running with sudo:

sudo npm run test:integration

Increase test timeout in jest.integration.config.js:

testTimeout: 300000, // 5 minutes

Run cleanup before tests:

./scripts/ci/cleanup.sh
npm run test:integration

Pre-pull Docker images:

docker pull curlimages/curl:latest
docker pull alpine:latest
docker pull dannydirect/tinyproxy:latest

Each test file follows a consistent structure:

/// <reference path="../jest-custom-matchers.d.ts" />

import { describe, test, expect, beforeAll, afterAll } from '@jest/globals';
import { createRunner, AwfRunner } from '../fixtures/awf-runner';
import { cleanup } from '../fixtures/cleanup';

describe('Feature Name', () => {
  let runner: AwfRunner;

  beforeAll(async () => {
    await cleanup(false);  // Clean up before tests
    runner = createRunner();
  });

  afterAll(async () => {
    await cleanup(false);  // Clean up after tests
  });

  test('should do something', async () => {
    const result = await runner.runWithSudo('command', {
      allowDomains: ['github.com'],
      logLevel: 'debug',
      timeout: 60000,
    });

    expect(result).toSucceed();
  }, 120000);  // Set individual test timeout
});

// Check success/failure
expect(result).toSucceed();
expect(result).toFail();

// Check specific exit code
expect(result).toExitWithCode(0);
expect(result).toExitWithCode(42);

// Check timeout
expect(result).toTimeout();

• Set reasonable timeouts for each test (typically 120000ms for integration tests)
• Use --max-time with curl to prevent indefinite hangs
• Set timeout in runner options

• Always run cleanup(false) in beforeAll and afterAll
• Use keepContainers: true only when needed for log inspection
• Clean up manually created files in afterEach

• Use explicit timeouts with network commands
• Don't depend on timing-sensitive conditions
• Use || true or error handling for expected failures
• Test for specific exit codes, not just success/failure

describe('Feature Category', () => {
  describe('Subsection A', () => {
    test('scenario 1', ...);
    test('scenario 2', ...);
  });

  describe('Subsection B', () => {
    test('scenario 3', ...);
    test('scenario 4', ...);
  });
});