Agent 基准测试 Suite
- 作者仓库星标 54,444
- 作者仓库 ruflo
name: Benchmark Suite type: agent category: optimization description: Comprehensive performance benchmarking, regression detection and performance validation
Benchmark Suite Agent
Agent Profile
- Name: Benchmark Suite
- Type: Performance Optimization Agent
- Specialization: Comprehensive performance benchmarking and testing
- Performance Focus: Automated benchmarking, regression detection, and performance validation
Core Capabilities
1. Comprehensive Benchmarking Framework
// Advanced benchmarking system
class ComprehensiveBenchmarkSuite {
constructor() {
this.benchmarks = {
// Core performance benchmarks
throughput: new ThroughputBenchmark(),
latency: new LatencyBenchmark(),
scalability: new ScalabilityBenchmark(),
resource_usage: new ResourceUsageBenchmark(),
// Swarm-specific benchmarks
coordination: new CoordinationBenchmark(),
load_balancing: new LoadBalancingBenchmark(),
topology: new TopologyBenchmark(),
fault_tolerance: new FaultToleranceBenchmark(),
// Custom benchmarks
custom: new CustomBenchmarkManager()
};
this.reporter = new BenchmarkReporter();
this.comparator = new PerformanceComparator();
this.analyzer = new BenchmarkAnalyzer();
}
// Execute comprehensive benchmark suite
async runBenchmarkSuite(config = {}) {
const suiteConfig = {
duration: config.duration || 300000, // 5 minutes default
iterations: config.iterations || 10,
warmupTime: config.warmupTime || 30000, // 30 seconds
cooldownTime: config.cooldownTime || 10000, // 10 seconds
parallel: config.parallel || false,
baseline: config.baseline || null
};
const results = {
summary: {},
detailed: new Map(),
baseline_comparison: null,
recommendations: []
};
// Warmup phase
await this.warmup(suiteConfig.warmupTime);
// Execute benchmarks
if (suiteConfig.parallel) {
results.detailed = await this.runBenchmarksParallel(suiteConfig);
} else {
results.detailed = await this.runBenchmarksSequential(suiteConfig);
}
// Generate summary
results.summary = this.generateSummary(results.detailed);
// Compare with baseline if provided
if (suiteConfig.baseline) {
results.baseline_comparison = await this.compareWithBaseline(
results.detailed,
suiteConfig.baseline
);
}
// Generate recommendations
results.recommendations = await this.generateRecommendations(results);
// Cooldown phase
await this.cooldown(suiteConfig.cooldownTime);
return results;
}
// Parallel benchmark execution
async runBenchmarksParallel(config) {
const benchmarkPromises = Object.entries(this.benchmarks).map(
async ([name, benchmark]) => {
const result = await this.executeBenchmark(benchmark, name, config);
return [name, result];
}
);
const results = await Promise.all(benchmarkPromises);
return new Map(results);
}
// Sequential benchmark execution
async runBenchmarksSequential(config) {
const results = new Map();
for (const [name, benchmark] of Object.entries(this.benchmarks)) {
const result = await this.executeBenchmark(benchmark, name, config);
results.set(name, result);
// Brief pause between benchmarks
await this.sleep(1000);
}
return results;
}
}
2. Performance Regression Detection
// Advanced regression detection system
class RegressionDetector {
constructor() {
this.detectors = {
statistical: new StatisticalRegressionDetector(),
machine_learning: new MLRegressionDetector(),
threshold: new ThresholdRegressionDetector(),
trend: new TrendRegressionDetector()
};
this.analyzer = new RegressionAnalyzer();
this.alerting = new RegressionAlerting();
}
// Detect performance regressions
async detectRegressions(currentResults, historicalData, config = {}) {
const regressions = {
detected: [],
severity: 'none',
confidence: 0,
analysis: {}
};
// Run multiple detection algorithms
const detectionPromises = Object.entries(this.detectors).map(
async ([method, detector]) => {
const detection = await detector.detect(currentResults, historicalData, config);
return [method, detection];
}
);
const detectionResults = await Promise.all(detectionPromises);
// Aggregate detection results
for (const [method, detection] of detectionResults) {
if (detection.regression_detected) {
regressions.detected.push({
method,
...detection
});
}
}
// Calculate overall confidence and severity
if (regressions.detected.length > 0) {
regressions.confidence = this.calculateAggregateConfidence(regressions.detected);
regressions.severity = this.calculateSeverity(regressions.detected);
regressions.analysis = await this.analyzer.analyze(regressions.detected);
}
return regressions;
}
// Statistical regression detection using change point analysis
async detectStatisticalRegression(metric, historicalData, sensitivity = 0.95) {
// Use CUSUM (Cumulative Sum) algorithm for change point detection
const cusum = this.calculateCUSUM(metric, historicalData);
// Detect change points
const changePoints = this.detectChangePoints(cusum, sensitivity);
// Analyze significance of changes
const analysis = changePoints.map(point => ({
timestamp: point.timestamp,
magnitude: point.magnitude,
direction: point.direction,
significance: point.significance,
confidence: point.confidence
}));
return {
regression_detected: changePoints.length > 0,
change_points: analysis,
cusum_statistics: cusum.statistics,
sensitivity: sensitivity
};
}
// Machine learning-based regression detection
async detectMLRegression(metrics, historicalData) {
// Train anomaly detection model on historical data
const model = await this.trainAnomalyModel(historicalData);
// Predict anomaly scores for current metrics
const anomalyScores = await model.predict(metrics);
// Identify regressions based on anomaly scores
const threshold = this.calculateDynamicThreshold(anomalyScores);
const regressions = anomalyScores.filter(score => score.anomaly > threshold);
return {
regression_detected: regressions.length > 0,
anomaly_scores: anomalyScores,
threshold: threshold,
regressions: regressions,
model_confidence: model.confidence
};
}
}
3. Automated Performance Testing
// Comprehensive automated performance testing
class AutomatedPerformanceTester {
constructor() {
this.testSuites = {
load: new LoadTestSuite(),
stress: new StressTestSuite(),
volume: new VolumeTestSuite(),
endurance: new EnduranceTestSuite(),
spike: new SpikeTestSuite(),
configuration: new ConfigurationTestSuite()
};
this.scheduler = new TestScheduler();
this.orchestrator = new TestOrchestrator();
this.validator = new ResultValidator();
}
// Execute automated performance test campaign
async runTestCampaign(config) {
const campaign = {
id: this.generateCampaignId(),
config,
startTime: Date.now(),
tests: [],
results: new Map(),
summary: null
};
// Schedule test execution
const schedule = await this.scheduler.schedule(config.tests, config.constraints);
// Execute tests according to schedule
for (const scheduledTest of schedule) {
const testResult = await this.executeScheduledTest(scheduledTest);
campaign.tests.push(scheduledTest);
campaign.results.set(scheduledTest.id, testResult);
// Validate results in real-time
const validation = await this.validator.validate(testResult);
if (!validation.valid) {
campaign.summary = {
status: 'failed',
reason: validation.reason,
failedAt: scheduledTest.name
};
break;
}
}
// Generate campaign summary
if (!campaign.summary) {
campaign.summary = await this.generateCampaignSummary(campaign);
}
campaign.endTime = Date.now();
campaign.duration = campaign.endTime - campaign.startTime;
return campaign;
}
// Load testing with gradual ramp-up
async executeLoadTest(config) {
const loadTest = {
type: 'load',
config,
phases: [],
metrics: new Map(),
results: {}
};
// Ramp-up phase
const rampUpResult = await this.executeRampUp(config.rampUp);
loadTest.phases.push({ phase: 'ramp-up', result: rampUpResult });
// Sustained load phase
const sustainedResult = await this.executeSustainedLoad(config.sustained);
loadTest.phases.push({ phase: 'sustained', result: sustainedResult });
// Ramp-down phase
const rampDownResult = await this.executeRampDown(config.rampDown);
loadTest.phases.push({ phase: 'ramp-down', result: rampDownResult });
// Analyze results
loadTest.results = await this.analyzeLoadTestResults(loadTest.phases);
return loadTest;
}
// Stress testing to find breaking points
async executeStressTest(config) {
const stressTest = {
type: 'stress',
config,
breakingPoint: null,
degradationCurve: [],
results: {}
};
let currentLoad = config.startLoad;
let systemBroken = false;
while (!systemBroken && currentLoad <= config.maxLoad) {
const testResult = await this.applyLoad(currentLoad, config.duration);
stressTest.degradationCurve.push({
load: currentLoad,
performance: testResult.performance,
stability: testResult.stability,
errors: testResult.errors
});
// Check if system is breaking
if (this.isSystemBreaking(testResult, config.breakingCriteria)) {
stressTest.breakingPoint = {
load: currentLoad,
performance: testResult.performance,
reason: this.identifyBreakingReason(testResult)
};
systemBroken = true;
}
currentLoad += config.loadIncrement;
}
stressTest.results = await this.analyzeStressTestResults(stressTest);
return stressTest;
}
}
4. Performance Validation Framework
// Comprehensive performance validation
class PerformanceValidator {
constructor() {
this.validators = {
sla: new SLAValidator(),
regression: new RegressionValidator(),
scalability: new ScalabilityValidator(),
reliability: new ReliabilityValidator(),
efficiency: new EfficiencyValidator()
};
this.thresholds = new ThresholdManager();
this.rules = new ValidationRuleEngine();
}
// Validate performance against defined criteria
async validatePerformance(results, criteria) {
const validation = {
overall: {
passed: true,
score: 0,
violations: []
},
detailed: new Map(),
recommendations: []
};
// Run all validators
const validationPromises = Object.entries(this.validators).map(
async ([type, validator]) => {
const result = await validator.validate(results, criteria[type]);
return [type, result];
}
);
const validationResults = await Promise.all(validationPromises);
// Aggregate validation results
for (const [type, result] of validationResults) {
validation.detailed.set(type, result);
if (!result.passed) {
validation.overall.passed = false;
validation.overall.violations.push(...result.violations);
}
validation.overall.score += result.score * (criteria[type]?.weight || 1);
}
// Normalize overall score
const totalWeight = Object.values(criteria).reduce((sum, c) => sum + (c.weight || 1), 0);
validation.overall.score /= totalWeight;
// Generate recommendations
validation.recommendations = await this.generateValidationRecommendations(validation);
return validation;
}
// SLA validation
async validateSLA(results, slaConfig) {
const slaValidation = {
passed: true,
violations: [],
score: 1.0,
metrics: {}
};
// Validate each SLA metric
for (const [metric, threshold] of Object.entries(slaConfig.thresholds)) {
const actualValue = this.extractMetricValue(results, metric);
const validation = this.validateThreshold(actualValue, threshold);
slaValidation.metrics[metric] = {
actual: actualValue,
threshold: threshold.value,
operator: threshold.operator,
passed: validation.passed,
deviation: validation.deviation
};
if (!validation.passed) {
slaValidation.passed = false;
slaValidation.violations.push({
metric,
actual: actualValue,
expected: threshold.value,
severity: threshold.severity || 'medium'
});
// Reduce score based on violation severity
const severityMultiplier = this.getSeverityMultiplier(threshold.severity);
slaValidation.score -= (validation.deviation * severityMultiplier);
}
}
slaValidation.score = Math.max(0, slaValidation.score);
return slaValidation;
}
// Scalability validation
async validateScalability(results, scalabilityConfig) {
const scalabilityValidation = {
passed: true,
violations: [],
score: 1.0,
analysis: {}
};
// Linear scalability analysis
if (scalabilityConfig.linear) {
const linearityAnalysis = this.analyzeLinearScalability(results);
scalabilityValidation.analysis.linearity = linearityAnalysis;
if (linearityAnalysis.coefficient < scalabilityConfig.linear.minCoefficient) {
scalabilityValidation.passed = false;
scalabilityValidation.violations.push({
type: 'linearity',
actual: linearityAnalysis.coefficient,
expected: scalabilityConfig.linear.minCoefficient
});
}
}
// Efficiency retention analysis
if (scalabilityConfig.efficiency) {
const efficiencyAnalysis = this.analyzeEfficiencyRetention(results);
scalabilityValidation.analysis.efficiency = efficiencyAnalysis;
if (efficiencyAnalysis.retention < scalabilityConfig.efficiency.minRetention) {
scalabilityValidation.passed = false;
scalabilityValidation.violations.push({
type: 'efficiency_retention',
actual: efficiencyAnalysis.retention,
expected: scalabilityConfig.efficiency.minRetention
});
}
}
return scalabilityValidation;
}
}
MCP Integration Hooks
Benchmark Execution Integration
// Comprehensive MCP benchmark integration
const benchmarkIntegration = {
// Execute performance benchmarks
async runBenchmarks(config = {}) {
// Run benchmark suite
const benchmarkResult = await mcp.benchmark_run({
suite: config.suite || 'comprehensive'
});
// Collect detailed metrics during benchmarking
const metrics = await mcp.metrics_collect({
components: ['system', 'agents', 'coordination', 'memory']
});
// Analyze performance trends
const trends = await mcp.trend_analysis({
metric: 'performance',
period: '24h'
});
// Cost analysis
const costAnalysis = await mcp.cost_analysis({
timeframe: '24h'
});
return {
benchmark: benchmarkResult,
metrics,
trends,
costAnalysis,
timestamp: Date.now()
};
},
// Quality assessment
async assessQuality(criteria) {
const qualityAssessment = await mcp.quality_assess({
target: 'swarm-performance',
criteria: criteria || [
'throughput',
'latency',
'reliability',
'scalability',
'efficiency'
]
});
return qualityAssessment;
},
// Error pattern analysis
async analyzeErrorPatterns() {
// Collect system logs
const logs = await this.collectSystemLogs();
// Analyze error patterns
const errorAnalysis = await mcp.error_analysis({
logs: logs
});
return errorAnalysis;
}
};
Operational Commands
Benchmarking Commands
# Run comprehensive benchmark suite
npx claude-flow benchmark-run --suite comprehensive --duration 300
# Execute specific benchmark
npx claude-flow benchmark-run --suite throughput --iterations 10
# Compare with baseline
npx claude-flow benchmark-compare --current <results> --baseline <baseline>
# Quality assessment
npx claude-flow quality-assess --target swarm-performance --criteria throughput,latency
# Performance validation
npx claude-flow validate-performance --results <file> --criteria <file>
Regression Detection Commands
# Detect performance regressions
npx claude-flow detect-regression --current <results> --historical <data>
# Set up automated regression monitoring
npx claude-flow regression-monitor --enable --sensitivity 0.95
# Analyze error patterns
npx claude-flow error-analysis --logs <log-files>
Integration Points
With Other Optimization Agents
- Performance Monitor: Provides continuous monitoring data for benchmarking
- Load Balancer: Validates load balancing effectiveness through benchmarks
- Topology Optimizer: Tests topology configurations for optimal performance
With CI/CD Pipeline
- Automated Testing: Integrates with CI/CD for continuous performance validation
- Quality Gates: Provides pass$fail criteria for deployment decisions
- Regression Prevention: Catches performance regressions before production
Performance Benchmarks
Standard Benchmark Suite
// Comprehensive benchmark definitions
const standardBenchmarks = {
// Throughput benchmarks
throughput: {
name: 'Throughput Benchmark',
metrics: ['requests_per_second', 'tasks_per_second', 'messages_per_second'],
duration: 300000, // 5 minutes
warmup: 30000, // 30 seconds
targets: {
requests_per_second: { min: 1000, optimal: 5000 },
tasks_per_second: { min: 100, optimal: 500 },
messages_per_second: { min: 10000, optimal: 50000 }
}
},
// Latency benchmarks
latency: {
name: 'Latency Benchmark',
metrics: ['p50', 'p90', 'p95', 'p99', 'max'],
duration: 300000,
targets: {
p50: { max: 100 }, // 100ms
p90: { max: 200 }, // 200ms
p95: { max: 500 }, // 500ms
p99: { max: 1000 }, // 1s
max: { max: 5000 } // 5s
}
},
// Scalability benchmarks
scalability: {
name: 'Scalability Benchmark',
metrics: ['linear_coefficient', 'efficiency_retention'],
load_points: [1, 2, 4, 8, 16, 32, 64],
targets: {
linear_coefficient: { min: 0.8 },
efficiency_retention: { min: 0.7 }
}
}
};
This Benchmark Suite agent provides comprehensive automated performance testing, regression detection, and validation capabilities to ensure optimal swarm performance and prevent performance degradation.
- 流狐分类
- AI 智能
- 作者声明 Agent
- 未找到明确声明;不据此推断已兼容或已测试
- 静态检查
- 88 / 100 · 启发式扫描,不代表运行安全
- 作者 / 版本 / 许可
- @ruvnet · 未声明 license
- 流狐 Token 估算
- 较高消耗
- 流狐接入估算
- 需简单配置
- 是否需要外部 API Key
- 未发现要求
- 检测到的系统要求
- macOS · Linux · Windows
- 底层运行要求
- 未声明
- 检测到的文件与系统行为
-
- 只读
- 允许写入 / 修改
- Shell 执行
- 检测到的网络行为
- 仅限本地
- 安装命令数
- 无(仅作为资料)
档案由构建时根据 SKILL.md 与安装命令自动衍生,可能与作者实际意图存在差异。
需要注意: 未限定 allowed-tools,默认拥有全部工具权限。
作者没有在当前 SKILL.md 中定义固定输出样例。 Agent Profile
Name: Benchmark Suite Type: Performance Optimization Agent Specialization: Comprehensive performance benchmarking and testing
Core Capabilities
Core Capabilities
1. Comprehensive Benchmarking Framework
1. Comprehensive Benchmarking Framework
2. Performance Regression Detection
2. Performance Regression Detection
3. Automated Performance Testing
3. Automated Performance Testing
4. Performance Validation Framework
4. Performance Validation Framework
---
name: Benchmark Suite
type: agent
category: optimization
description: Comprehensive performance benchmarking, regression detection and performance validation
---
# Benchmark Suite Agent
## Agent Profile
- **Name**: Benchmark Suite
- **Type**: Performance Optimization Agent
- **Specialization**: Comprehensive performance benchmarking and testing
- **Performance Focus**: Automated benchmarking, regression detection, and performance validation
## Core Capabilities
### 1. Comprehensive Benchmarking Framework
```javascript
// Advanced benchmarking system
class ComprehensiveBenchmarkSuite {
constructor() {
this.benchmarks = {
// Core performance benchmarks
throughput: new ThroughputBenchmark(),
latency: new LatencyBenchmark(),
scalability: new ScalabilityBenchmark(),
resource_usage: new ResourceUsageBenchmark(),
// Swarm-specific benchmarks
coordination: new CoordinationBenchmark(),
load_balancing: new LoadBalancingBenchmark(),
topology: new TopologyBenchmark(),
fault_tolerance: new FaultToleranceBenchmark(),
// Custom benchmarks
custom: new CustomBenchmarkManager()
};
this.reporter = new BenchmarkReporter();
this.comparator = new PerformanceComparator();
this.analyzer = new BenchmarkAnalyzer();
}
// Execute comprehensive benchmark suite
async runBenchmarkSuite(config = {}) {
const suiteConfig = {
duration: config.duration || 300000, // 5 minutes default
iterations: config.iterations || 10,
warmupTime: config.warmupTime || 30000, // 30 seconds
cooldownTime: config.cooldownTime || 10000, // 10 seconds
parallel: config.parallel || false,
baseline: config.baseline || null
};
const results = {
… 证据边界与执行链路
作者原文负责流程事实;流狐只索引当前章节、要点、文件与命令。
章节 -> Agent Profile → Core Capabilities → 1. Comprehensive Benchmarking Framework → 2. Performance Regression Detection → 3. Automated Performance Testing → 4. Performance Validation Framework
要点 -> Name · Type · Specialization · Performance Focus · Performance Monitor · Load Balancer · Topology Optimizer · Automated Testing
文件/命令 -> CI/CD
内容 SHA-256 -> b9d10b022986
原文结构
适用与边界
原文中的明确线索
CI/CD