test-server.ts•10 kB
/**
* Comprehensive Test Suite for Claude Vision MCP Server
* Tests all three tools: think, describe_image, and analyze_image
*/
import Anthropic from "@anthropic-ai/sdk";
import fs from "fs";
import path from "path";
import https from "https";
// Load environment variables
const API_KEY = process.env.ANTHROPIC_API_KEY || '';
const MODEL = "claude-sonnet-4-20250514";
if (!API_KEY || API_KEY === 'your-anthropic-api-key-here') {
console.error("❌ ERROR: Please set ANTHROPIC_API_KEY environment variable");
console.error(" Get your key from: https://console.anthropic.com/");
process.exit(1);
}
const anthropic = new Anthropic({ apiKey: API_KEY });
// Test results storage
interface TestResult {
name: string;
passed: boolean;
duration: number;
error?: string;
output?: string;
}
const results: TestResult[] = [];
// Helper function to measure execution time
async function measureTest(name: string, testFn: () => Promise<void>): Promise<void> {
console.log(`\n${'='.repeat(60)}`);
console.log(`🧪 Running: ${name}`);
console.log('='.repeat(60));
const startTime = Date.now();
try {
await testFn();
const duration = Date.now() - startTime;
results.push({ name, passed: true, duration });
console.log(`✅ PASSED in ${duration}ms (${(duration / 1000).toFixed(2)}s)`);
} catch (error: any) {
const duration = Date.now() - startTime;
results.push({ name, passed: false, duration, error: error.message });
console.error(`❌ FAILED: ${error.message}`);
}
}
// Download test image
async function downloadTestImage(url: string, outputPath: string): Promise<void> {
return new Promise((resolve, reject) => {
const file = fs.createWriteStream(outputPath);
https.get(url, (response) => {
response.pipe(file);
file.on('finish', () => {
file.close();
resolve();
});
}).on('error', (err) => {
fs.unlink(outputPath, () => {});
reject(err);
});
});
}
/**
* TEST 1: Deep Think Tool - Text Analysis
*/
async function testThinkTool(): Promise<void> {
console.log("\n📝 Testing 'think' tool for deep analysis...\n");
const query = `I'm building a real-time chat application that needs to support 100,000 concurrent users.
Should I use WebSockets, Server-Sent Events (SSE), or HTTP polling?
Consider scalability, browser compatibility, and infrastructure costs.`;
const context = "Building a SaaS chat platform with real-time requirements";
console.log("Query:", query.substring(0, 100) + "...");
console.log("\nCalling Claude API...");
const response = await anthropic.messages.create({
model: MODEL,
max_tokens: 4096,
system: `You are an expert analytical assistant. Provide deep, actionable insights with structured analysis.`,
messages: [
{
role: "user",
content: `Context: ${context}\n\nQuery: ${query}`,
},
],
});
const analysisText = response.content
.filter((block) => block.type === "text")
.map((block) => (block as { type: "text"; text: string }).text)
.join("\n");
console.log("\n" + "─".repeat(60));
console.log("📊 RESULT:");
console.log("─".repeat(60));
console.log(analysisText.substring(0, 500) + "...\n");
console.log(`Token Usage: ${response.usage.input_tokens} in / ${response.usage.output_tokens} out`);
if (analysisText.length < 100) {
throw new Error("Analysis too short - expected detailed response");
}
}
/**
* TEST 2: Quick Image Description - URL Analysis
*/
async function testDescribeImageTool(): Promise<void> {
console.log("\n🖼️ Testing 'describe_image' tool with URL...\n");
const imageUrl = "https://picsum.photos/seed/mcp-test-123/800/600";
console.log("Image URL:", imageUrl);
console.log("\nCalling Claude Vision API...");
const response = await anthropic.messages.create({
model: MODEL,
max_tokens: 2048,
messages: [
{
role: "user",
content: [
{
type: "image",
source: {
type: "url",
url: imageUrl,
},
},
{
type: "text",
text: "Please provide a comprehensive description of this image, including key elements, composition, colors, mood, and any notable details.",
},
],
},
],
});
const description = response.content
.filter((block) => block.type === "text")
.map((block) => (block as { type: "text"; text: string }).text)
.join("\n");
console.log("\n" + "─".repeat(60));
console.log("📊 RESULT:");
console.log("─".repeat(60));
console.log(description.substring(0, 500) + "...\n");
console.log(`Token Usage: ${response.usage.input_tokens} in / ${response.usage.output_tokens} out`);
if (description.length < 50) {
throw new Error("Description too short - expected detailed response");
}
}
/**
* TEST 3: Multi-Perspective Image Analysis - Local File
*/
async function testAnalyzeImageTool(): Promise<void> {
console.log("\n🔍 Testing 'analyze_image' tool with local file...\n");
// Download a test image
const imageUrl = "https://images.unsplash.com/photo-1557821552-17105176677c?w=800";
const imagePath = path.join(process.cwd(), "claude_vision", "test-image.jpg");
console.log("Downloading test image...");
await downloadTestImage(imageUrl, imagePath);
console.log("✓ Image downloaded to:", imagePath);
// Read and encode image
const imageBuffer = fs.readFileSync(imagePath);
const base64Image = imageBuffer.toString("base64");
const iterations = 3; // Using 3 for faster testing
const analysisResults: string[] = [];
const perspectives = [
"Overall atmosphere, mood, and emotional tone",
"Visual composition, layout, and design elements",
"Colors, lighting, and photographic techniques",
];
console.log(`\nPerforming ${iterations} iterations of analysis...\n`);
for (let i = 0; i < iterations; i++) {
const perspective = perspectives[i];
console.log(` Iteration ${i + 1}/${iterations}: ${perspective}`);
const response = await anthropic.messages.create({
model: MODEL,
max_tokens: 1024,
system: `You are an expert image analyst. Focus specifically on: ${perspective}`,
messages: [
{
role: "user",
content: [
{
type: "image",
source: {
type: "base64",
media_type: "image/jpeg",
data: base64Image,
},
},
{
type: "text",
text: `Analyze this image focusing on: ${perspective}`,
},
],
},
],
});
const analysisText = response.content
.filter((block) => block.type === "text")
.map((block) => (block as { type: "text"; text: string }).text)
.join("\n");
analysisResults.push(`**Perspective ${i + 1}:** ${analysisText.substring(0, 200)}...`);
}
console.log("\n" + "─".repeat(60));
console.log("📊 MULTI-PERSPECTIVE ANALYSIS RESULTS:");
console.log("─".repeat(60));
analysisResults.forEach((result, idx) => {
console.log(`\n${idx + 1}. ${result}`);
});
console.log("\n✓ All iterations completed successfully!");
if (analysisResults.length !== iterations) {
throw new Error(`Expected ${iterations} iterations, got ${analysisResults.length}`);
}
}
/**
* Print Test Summary
*/
function printSummary(): void {
console.log("\n\n");
console.log("═".repeat(70));
console.log(" 📊 TEST SUMMARY");
console.log("═".repeat(70));
const passed = results.filter(r => r.passed).length;
const failed = results.filter(r => !r.passed).length;
const total = results.length;
console.log(`\nTotal Tests: ${total} | Passed: ✅ ${passed} | Failed: ❌ ${failed}\n`);
results.forEach((result, idx) => {
const status = result.passed ? "✅ PASS" : "❌ FAIL";
const duration = `${(result.duration / 1000).toFixed(2)}s`;
console.log(`${idx + 1}. ${status} - ${result.name} (${duration})`);
if (result.error) {
console.log(` Error: ${result.error}`);
}
});
console.log("\n" + "═".repeat(70));
const totalDuration = results.reduce((sum, r) => sum + r.duration, 0);
console.log(`Total Duration: ${(totalDuration / 1000).toFixed(2)}s`);
console.log(`Success Rate: ${((passed / total) * 100).toFixed(1)}%`);
console.log("═".repeat(70) + "\n");
if (failed > 0) {
console.error("⚠️ Some tests failed. Check the output above for details.\n");
process.exit(1);
} else {
console.log("🎉 All tests passed successfully!\n");
console.log("Your Claude Vision MCP Server is working perfectly! ✨\n");
}
}
/**
* Main Test Runner
*/
async function runTests(): Promise<void> {
console.log("\n");
console.log("╔" + "═".repeat(68) + "╗");
console.log("║" + " ".repeat(68) + "║");
console.log("║" + " 🧪 Claude Vision MCP Server - Comprehensive Test Suite ".padEnd(68) + "║");
console.log("║" + " ".repeat(68) + "║");
console.log("╚" + "═".repeat(68) + "╝");
console.log("\n");
console.log("🔑 API Key: " + API_KEY.substring(0, 20) + "..." + API_KEY.substring(API_KEY.length - 10));
console.log("🤖 Model: " + MODEL);
console.log("\n");
// Ensure claude_vision directory exists
const visionDir = path.join(process.cwd(), "claude_vision");
if (!fs.existsSync(visionDir)) {
fs.mkdirSync(visionDir, { recursive: true });
}
// Run all tests
await measureTest("Test 1: Deep Think Tool (Text Analysis)", testThinkTool);
await measureTest("Test 2: Quick Image Description (URL)", testDescribeImageTool);
await measureTest("Test 3: Multi-Perspective Image Analysis (Local File)", testAnalyzeImageTool);
// Print summary
printSummary();
}
// Execute tests
runTests().catch((error) => {
console.error("\n❌ Fatal error running tests:", error);
process.exit(1);
});