performance-analysis

Bottleneck detection in Claude Flow swarms identifies slowdowns across communication, processing, memory, and network layers. Run `npx claude-flow bottleneck detect` to automatically pinpoint which components are causing delays, then review the detailed analysis to understand where optimization is needed.

Performance profiling in multi-agent systems exposes inter-agent communication delays, inefficient memory access, compute bottlenecks, and network latency. Real-time monitoring and historical analysis show exactly where agents are slowing down, enabling targeted optimization across your swarm's coordination and processing layers.

Generate comprehensive performance reports by running `npx claude-flow analysis performance-report --format html --include-metrics`. Export supports multiple formats and includes detailed metrics showing bottleneck locations, severity, and swarm behavior, making it easy to share findings and track improvements over time.

Yes, automatic fixing is available with threshold-driven correction. Run `npx claude-flow bottleneck detect --fix --threshold 15` to detect bottlenecks and apply recommended fixes simultaneously. The system implements actionable optimizations like topology changes and caching strategies to boost swarm efficiency.

Optimization recommendations are AI-powered and tailored to detected issues. Common suggestions include switching to hierarchical topologies to reduce communication overhead, enabling memory caching to improve access speed, and rebalancing compute distribution. Recommendations target the specific bottleneck types found in your swarm.

Yes, Claude Flow is a required dependency for this Skill. The performance analysis tools work directly with Claude Flow swarms through CLI commands, so you must have Claude Flow set up and your swarm initialized before running bottleneck detection, profiling, or optimization analysis.