Анализ производительности и узких мест Claude Flow

testing

performance-analysis is a Claude Code skill that provides comprehensive bottleneck detection, profiling, and optimization recommendations for Claude Flow swarms. Using `npx claude-flow bottleneck detect` and `npx claude-flow analysis performance-report`, it examines agent communication latency, CPU and memory usage, parallel execution efficiency, cache hit rates, and MCP network delays across configurable time ranges from one hour to 30 days. Reports are produced in JSON, HTML, or Markdown; the `--fix` flag applies automatic fixes such as enabling smart caching, optimizing message routing, and rebalancing agent priorities. Results include quantified impact estimates — for example, switching to a hierarchical topology for an estimated 40% improvement — making this skill practical for teams diagnosing performance regressions in multi-agent pipelines or tuning swarm throughput at scale.