NCP - Natural Context Provider

#!/usr/bin/env ts-node /** * Automated test for SDK-based MCP server * Uses MCP SDK Client to connect and test the server */ import { Client } from '@modelcontextprotocol/sdk/client/index.js'; import { StdioClientTransport } from '@modelcontextprotocol/sdk/client/stdio.js'; import { spawn } from 'child_process'; import { fileURLToPath } from 'url'; import { dirname, join } from 'path'; const __filename = fileURLToPath(import.meta.url); const __dirname = dirname(__filename); interface TestResult { name: string; passed: boolean; error?: string; details?: any; } const results: TestResult[] = []; function logTest(name: string, passed: boolean, details?: any, error?: string) { results.push({ name, passed, error, details }); const icon = passed ? '✅' : '❌'; console.log(`${icon} ${name}`); if (details) { console.log(` Details: ${JSON.stringify(details, null, 2)}`); } if (error) { console.log(` Error: ${error}`); } } async function runTests() { console.log('🧪 Starting SDK Server Tests\n'); let client: Client | null = null; let transport: StdioClientTransport | null = null; try { // Step 1: Create transport to connect to our server console.log('📡 Connecting to NCP SDK server...'); const serverPath = join(__dirname, 'dist', 'index-mcp.js'); transport = new StdioClientTransport({ command: 'node', args: [serverPath], env: { ...process.env } }); // Step 2: Create client client = new Client( { name: 'test-client', version: '1.0.0' }, { capabilities: {} } ); // Step 3: Connect await client.connect(transport); logTest('Server Connection', true, { message: 'Connected successfully' }); // Step 4: Get server info const serverInfo = client.getServerVersion(); logTest('Server Info', !!serverInfo, { name: serverInfo?.name, version: serverInfo?.version }); // Step 5: List tools console.log('\n🔧 Testing tools/list...'); const toolsResponse = await client.listTools(); const tools = toolsResponse.tools; logTest('Tools List Response', tools.length > 0, { toolCount: tools.length, toolNames: tools.map(t => t.name) }); // Step 6: Verify 'find' tool exists const findTool = tools.find(t => t.name === 'find'); logTest('Find Tool Available', !!findTool, { hasDescription: !!findTool?.description, hasSchema: !!findTool?.inputSchema }); // Step 7: Verify 'run' tool exists const runTool = tools.find(t => t.name === 'run'); logTest('Run Tool Available', !!runTool, { hasDescription: !!runTool?.description, hasSchema: !!runTool?.inputSchema }); // Step 8: Test calling 'find' tool (list mode) console.log('\n🔍 Testing find tool (list mode)...'); try { const findResult = await client.callTool({ name: 'find', arguments: {} }); const hasContent = findResult.content && findResult.content.length > 0; logTest('Find Tool Execution (list mode)', hasContent, { contentType: findResult.content[0]?.type, hasText: !!(findResult.content[0] as any)?.text }); } catch (error: any) { logTest('Find Tool Execution (list mode)', false, undefined, error.message); } // Step 9: Test calling 'find' tool (search mode) console.log('\n🔍 Testing find tool (search mode)...'); try { const searchResult = await client.callTool({ name: 'find', arguments: { description: 'file operations', limit: 3 } }); const hasContent = searchResult.content && searchResult.content.length > 0; logTest('Find Tool Execution (search mode)', hasContent, { contentType: searchResult.content[0]?.type, hasText: !!(searchResult.content[0] as any)?.text }); } catch (error: any) { logTest('Find Tool Execution (search mode)', false, undefined, error.message); } // Step 10: Test 'run' tool with dry_run console.log('\n🚀 Testing run tool (dry run)...'); try { const runResult = await client.callTool({ name: 'run', arguments: { tool: 'test:sample', parameters: { example: 'value' }, dry_run: true } }); const hasContent = runResult.content && runResult.content.length > 0; const text = (runResult.content[0] as any)?.text || ''; const isDryRun = text.includes('DRY RUN'); logTest('Run Tool Execution (dry run)', hasContent && isDryRun, { contentType: runResult.content[0]?.type, isDryRunPreview: isDryRun }); } catch (error: any) { logTest('Run Tool Execution (dry run)', false, undefined, error.message); } // Step 11: Test internal MCPs discovery (ncp:list, ncp:add, etc.) console.log('\n🔍 Testing internal MCPs discovery...'); try { // Wait a moment for initialization to complete await new Promise(resolve => setTimeout(resolve, 2000)); const ncpSearchResult = await client.callTool({ name: 'find', arguments: { description: 'add MCP server configuration', limit: 10 } }); const hasContent = ncpSearchResult.content && ncpSearchResult.content.length > 0; const text = (ncpSearchResult.content[0] as any)?.text || ''; const hasInternalTools = text.includes('ncp:') && (text.includes('add') || text.includes('list') || text.includes('import')); logTest('Internal MCPs Discovery', hasContent && hasInternalTools, { contentType: ncpSearchResult.content[0]?.type, foundInternalMCPs: hasInternalTools, textSample: text.substring(0, 300) }); } catch (error: any) { logTest('Internal MCPs Discovery', false, undefined, error.message); } // Step 12: Test calling internal MCP tool (ncp:list) console.log('\n🔧 Testing internal MCP execution (ncp:list)...'); try { const listResult = await client.callTool({ name: 'run', arguments: { tool: 'ncp:list', parameters: {} } }); const hasContent = listResult.content && listResult.content.length > 0; const text = (listResult.content[0] as any)?.text || ''; const isSuccess = !((listResult as any).isError) && (text.includes('MCP') || text.includes('profile')); logTest('Internal MCP Execution (ncp:list)', hasContent && isSuccess, { contentType: listResult.content[0]?.type, hasResponse: !!text, isError: !!(listResult as any).isError }); } catch (error: any) { logTest('Internal MCP Execution (ncp:list)', false, undefined, error.message); } // Step 13: Test error handling (invalid tool) console.log('\n⚠️ Testing error handling...'); try { const errorResult = await client.callTool({ name: 'nonexistent-tool', arguments: {} }); // SDK might return error in result instead of throwing const hasError = (errorResult as any).isError || (errorResult.content && errorResult.content[0] && (errorResult.content[0] as any).text?.includes('not found')); logTest('Error Handling (invalid tool)', hasError, { errorHandled: true, errorInResponse: !!(errorResult as any).isError }); } catch (error: any) { const hasErrorMessage = error.message && ( error.message.includes('not found') || error.message.includes('Method not found') ); logTest('Error Handling (invalid tool)', hasErrorMessage, { errorReceived: true, errorMessage: error.message }); } // Step 14: Cleanup console.log('\n🧹 Cleaning up...'); await client.close(); logTest('Client Disconnect', true); } catch (error: any) { console.error('\n❌ Test suite failed:', error.message); if (error.stack) { console.error('Stack:', error.stack); } if (client) { try { await client.close(); } catch (e) { // Ignore cleanup errors } } } // Print summary console.log('\n' + '='.repeat(50)); console.log('📊 Test Summary'); console.log('='.repeat(50)); const passed = results.filter(r => r.passed).length; const failed = results.filter(r => !r.passed).length; const total = results.length; console.log(`Total: ${total}`); console.log(`Passed: ${passed} ✅`); console.log(`Failed: ${failed} ${failed > 0 ? '❌' : ''}`); console.log(`Success Rate: ${Math.round((passed / total) * 100)}%`); if (failed > 0) { console.log('\n❌ Failed Tests:'); results.filter(r => !r.passed).forEach(r => { console.log(` - ${r.name}: ${r.error || 'Unknown error'}`); }); } // Exit with appropriate code process.exit(failed > 0 ? 1 : 0); } // Run tests runTests().catch(error => { console.error('Fatal error:', error); process.exit(1); });