SSH MCP Server

/** * Unified Broadcast+Store Test Suite * * This test suite validates the unified approach to terminal command echo and history storage, * ensuring that what gets broadcast to live browsers is exactly what gets stored in history. * * The goal is to eliminate the dual storage/broadcast logic that causes inconsistency between * live display and history replay. * * Target: Single unified function that broadcasts AND stores exactly the same content. */ import { JestTestUtilities } from './integration/terminal-history-framework/jest-test-utilities'; import { TestEnvironmentConfig } from './test-environment-config'; describe('Unified Broadcast+Store Behavior', () => { const testUtils = JestTestUtilities.setupJestEnvironment('unified-broadcast-store'); describe('MCP Command Echo Behavior', () => { it('should show prompt+command, then result, then final prompt for MCP commands', async () => { // ARRANGE - Test MCP command execution (source='claude') const username = TestEnvironmentConfig.getTestUsername(); const sshKeyPath = TestEnvironmentConfig.getTestSSHKeyPath(); const prompt = TestEnvironmentConfig.getExpectedPrompt(); const config = { preWebSocketCommands: [ `ssh_connect {"name": "mcp-command-test", "host": "localhost", "username": "${username}", "keyFilePath": "${sshKeyPath}"}`, 'ssh_exec {"sessionName": "mcp-command-test", "command": "whoami"}' // This is an MCP command (source='claude') ], postWebSocketCommands: [], workflowTimeout: 30000, sessionName: 'mcp-command-test' }; // ACT const result = await testUtils.runTerminalHistoryTest(config); // ASSERT - MCP commands should show: prompt+command → result → final prompt expect(result.success).toBe(true); const trailingPrompt = TestEnvironmentConfig.getExpectedTrailingPrompt(); const expectedOutput = `${prompt} whoami\r\n${username}\r\n${trailingPrompt}`; console.log('=== MCP COMMAND EXACT VALIDATION ==='); console.log('Expected:', JSON.stringify(expectedOutput)); console.log('Actual :', JSON.stringify(result.concatenatedResponses)); console.log('======================================='); // This test will FAIL initially because current logic creates double prompts expect(result.concatenatedResponses).toBe(expectedOutput); console.log('✅ MCP COMMAND ECHO VALIDATED - EXACT MATCH'); }); }); describe('Browser Command Echo Behavior', () => { it('should show command, then result, then final prompt for browser commands', async () => { // ARRANGE - Test browser command execution (source='user') via post-WebSocket const username = TestEnvironmentConfig.getTestUsername(); const sshKeyPath = TestEnvironmentConfig.getTestSSHKeyPath(); const config = { preWebSocketCommands: [ `ssh_connect {"name": "browser-command-test", "host": "localhost", "username": "${username}", "keyFilePath": "${sshKeyPath}"}` ], postWebSocketCommands: [ {initiator: 'browser', command: 'whoami'} // This simulates browser user typing ], workflowTimeout: 30000, sessionName: 'browser-command-test' }; // ACT const result = await testUtils.runTerminalHistoryTest(config); // ASSERT - Browser commands should show: command → result → final prompt (no initial prompt) expect(result.success).toBe(true); const trailingPrompt = TestEnvironmentConfig.getExpectedTrailingPrompt(); const expectedOutput = `whoami\r\n${username}\r\n${trailingPrompt}`; console.log('=== BROWSER COMMAND EXACT VALIDATION ==='); console.log('Expected:', JSON.stringify(expectedOutput)); console.log('Actual :', JSON.stringify(result.concatenatedResponses)); console.log('========================================'); // This test will likely FAIL initially due to current complex logic expect(result.concatenatedResponses).toBe(expectedOutput); console.log('✅ BROWSER COMMAND ECHO VALIDATED - EXACT MATCH'); }); }); describe('History Replay Consistency', () => { it('should replay stored history exactly as it was broadcast live', async () => { // ARRANGE - Execute multiple commands and verify replay consistency const username = TestEnvironmentConfig.getTestUsername(); const sshKeyPath = TestEnvironmentConfig.getTestSSHKeyPath(); const prompt = TestEnvironmentConfig.getExpectedPrompt(); const config = { preWebSocketCommands: [ `ssh_connect {"name": "history-consistency-test", "host": "localhost", "username": "${username}", "keyFilePath": "${sshKeyPath}"}`, 'ssh_exec {"sessionName": "history-consistency-test", "command": "echo test"}' ], postWebSocketCommands: [ {initiator: 'mcp-client', command: 'ssh_exec {"sessionName": "history-consistency-test", "command": "whoami"}'} ], workflowTimeout: 30000, sessionName: 'history-consistency-test' }; // ACT const result = await testUtils.runTerminalHistoryTest(config); // ASSERT - History should be exactly what was broadcast expect(result.success).toBe(true); // Unified behavior: MCP commands show prompt+command, browser commands just show command const trailingPrompt = TestEnvironmentConfig.getExpectedTrailingPrompt(); const expectedOutput = `${prompt} echo test\r\ntest\r\n${trailingPrompt}whoami\r\n${username}\r\n${trailingPrompt}`; console.log('=== HISTORY CONSISTENCY EXACT VALIDATION ==='); console.log('Expected:', JSON.stringify(expectedOutput)); console.log('Actual :', JSON.stringify(result.concatenatedResponses)); console.log('============================================'); // This test validates that stored history = broadcast content exactly expect(result.concatenatedResponses).toBe(expectedOutput); console.log('✅ HISTORY REPLAY CONSISTENCY VALIDATED'); }); }); describe('No Double Prompt Prevention', () => { it('should never show duplicate or concatenated prompts', async () => { // ARRANGE - Test scenario that previously caused double prompts const username = TestEnvironmentConfig.getTestUsername(); const sshKeyPath = TestEnvironmentConfig.getTestSSHKeyPath(); const prompt = TestEnvironmentConfig.getExpectedPrompt(); const config = { preWebSocketCommands: [ `ssh_connect {"name": "no-double-prompt-test", "host": "localhost", "username": "${username}", "keyFilePath": "${sshKeyPath}"}`, 'ssh_exec {"sessionName": "no-double-prompt-test", "command": "pwd"}' ], postWebSocketCommands: [], workflowTimeout: 30000, sessionName: 'no-double-prompt-test' }; // ACT const result = await testUtils.runTerminalHistoryTest(config); // ASSERT - No double prompts, clean single sequence expect(result.success).toBe(true); const homeDir = TestEnvironmentConfig.getTestHomeDirectory(); const trailingPrompt = TestEnvironmentConfig.getExpectedTrailingPrompt(); const expectedOutput = `${prompt} pwd\r\n${homeDir}\r\n${trailingPrompt}`; console.log('=== NO DOUBLE PROMPT EXACT VALIDATION ==='); console.log('Expected:', JSON.stringify(expectedOutput)); console.log('Actual :', JSON.stringify(result.concatenatedResponses)); console.log('========================================='); // Verify no double prompts exist in the output const doublePromptPattern = `${prompt}${prompt}`; expect(result.concatenatedResponses.includes(doublePromptPattern)).toBe(false); // Verify exact expected output expect(result.concatenatedResponses).toBe(expectedOutput); console.log('✅ NO DOUBLE PROMPTS DETECTED - CLEAN OUTPUT'); }); }); describe('Unified Function Behavior', () => { it('should use single unified broadcastAndStore function for all command types', async () => { // This test will verify that the implementation uses a single unified function // instead of separate broadcastToLiveListeners + storeInHistory calls // NOTE: This test primarily validates the internal implementation approach // The external behavior validation is covered by the other tests // ARRANGE - Simple command to test unified function usage const username = TestEnvironmentConfig.getTestUsername(); const sshKeyPath = TestEnvironmentConfig.getTestSSHKeyPath(); const config = { preWebSocketCommands: [ `ssh_connect {"name": "unified-function-test", "host": "localhost", "username": "${username}", "keyFilePath": "${sshKeyPath}"}`, 'ssh_exec {"sessionName": "unified-function-test", "command": "echo unified"}' ], postWebSocketCommands: [], workflowTimeout: 30000, sessionName: 'unified-function-test' }; // ACT const result = await testUtils.runTerminalHistoryTest(config); // ASSERT - Simple validation that unified approach works expect(result.success).toBe(true); expect(result.concatenatedResponses).toContain('unified'); expect(result.concatenatedResponses).toContain('\r\n'); console.log('✅ UNIFIED FUNCTION APPROACH VALIDATED'); console.log('📋 Output:', JSON.stringify(result.concatenatedResponses)); }); }); afterAll(() => { console.log('\n📊 UNIFIED BROADCAST+STORE TEST SUMMARY:'); console.log('🎯 Validates unified approach requirements:'); console.log(' ✓ MCP commands: prompt+command → result → final prompt'); console.log(' ✓ Browser commands: command → result → final prompt'); console.log(' ✓ History replay = exact broadcast content'); console.log(' ✓ No double prompts or concatenation'); console.log(' ✓ Single unified broadcastAndStore function'); console.log('\n🛠️ Expected Implementation Changes:'); console.log(' • Replace broadcastToLiveListeners + storeInHistory with unified function'); console.log(' • Eliminate complex conditional logic in broadcastCommandWithPrompt'); console.log(' • Store exactly what gets broadcast - no separate processing'); console.log('\n🚀 These tests will PASS after implementing unified approach!'); }); });