Network Performance
The Performance Analyzer: Network Performance agent is a specialized performance analyzer focused on network and HTTP bottlenecks. It finds code patterns where network usage is inefficient, causing slow page loads, high bandwidth costs, or unnecessary latency.
When to Use
Use this agent when:
- You need to identify HTTP waterfall patterns (sequential API calls)
- You want to check for missing request batching opportunities
- You're analyzing code for large or uncompressed API payloads
- You need to find excessive polling patterns
- You're looking for missing HTTP/2, keep-alive, or compression optimizations
How It Works
- Reads target code - Focuses on fetch/axios calls, API route handlers, polling mechanisms, server configuration, data transfer patterns
- Identifies patterns - Looks for sequential awaits, missing Promise.all, individual API calls instead of batching, over-fetching (full objects vs needed fields), polling at short intervals, missing compression
- Reports findings - Generates structured findings with specific locations, severity levels, latency impact, and remediation steps
- Provides context - Shows exact code and quantifies network performance improvement
Focus Areas
- HTTP waterfall: Sequential
await fetch()calls that could be parallelized withPromise.all - Missing request batching: Multiple individual API calls that could be combined into one batch request
- No compression: Missing gzip/brotli compression on server responses, uncompressed API payloads
- Large payloads: API responses returning full objects when only a few fields are needed (over-fetching)
- Excessive polling: Short polling intervals, polling when WebSocket/SSE would be more efficient
- Missing connection optimization: No HTTP/2, no keep-alive, no connection pooling
Tools Available
This agent has access to: Read, Glob, Grep
Example Analysis
Given this code:
const user = await fetch('/api/user');
const orders = await fetch('/api/orders');
const notifications = await fetch('/api/notifications');The Network Performance analyzer would identify:
Finding: Sequential API calls in HTTP waterfall
Location: pages/dashboard.ts:42
Severity: CRITICAL
Confidence: HIGH
Issue: Three independent API calls are made sequentially instead of in parallel. If each call takes ~300ms, the total latency is ~900ms instead of ~300ms.
Impact Estimate:
- Current: "3 sequential requests = 900ms total latency"
- Expected: "3 parallel requests = 300ms total latency"
- Savings: "~600ms per page load (67% improvement)"
Suggested Fix:
const [user, orders, notifications] = await Promise.all([
fetch('/api/user'),
fetch('/api/orders'),
fetch('/api/notifications')
]);Best Practices
- Parallelize independent API calls with
Promise.all() - Use batch endpoints to combine multiple queries into one request
- Implement HTTP compression middleware (gzip, brotli)
- Use HTTP/2 for connection multiplexing (most modern servers support this)
- Implement caching with proper Cache-Control headers
- Minimize payload size by selecting only needed fields (projections)
- Use pagination for large result sets
- Implement proper ETags and conditional requests
- Consider WebSocket for real-time data instead of polling
- Use GraphQL or JSON:API for flexible field selection
- Monitor with tools like WebPageTest or Lighthouse
Output Format
For each potential issue, the agent provides:
- Location: Exact file path and line number
- Severity: CRITICAL (major latency or bandwidth waste), HIGH (noticeable impact), MEDIUM (optimization), or LOW (minor improvement)
- Confidence: HIGH, MEDIUM, or LOW
- Category: HTTP Waterfall, Missing Batching, Over-Fetching, Excessive Polling, Missing Compression, or Missing Cache Headers
- Code: Relevant code snippet
- Issue: Clear explanation of network performance impact
- Impact Estimate: Current vs expected latency/bandwidth with quantified savings
- Remediation: Specific fix with code example
Example Usage
Task(
description: "Analyze network performance in React application",
prompt: "Review src/ for HTTP waterfall patterns, sequential API calls, and large payloads. Focus on page load and interaction critical paths.",
subagent_type: "agileflow-perf-analyzer-network"
)Related Agents
perf-analyzer-queries- Database query optimizationperf-analyzer-rendering- UI rendering performanceperf-analyzer-memory- Memory leaks and retentionperf-analyzer-bundle- Bundle size optimizationperf-analyzer-compute- CPU and compute efficiencyperf-analyzer-caching- Caching opportunitiesperf-analyzer-assets- Asset optimizationperf-consensus- Performance audit consensus coordinator