performance-analysis

Performance bottleneck detection identifies communication, processing, memory, and network constraints across your swarm automatically. Run the bottleneck detection command to scan operations in real time or analyze historical data, then review the generated report to pinpoint which agent interactions or resources are slowing execution.

Performance profiling detects communication delays between agents, processing inefficiencies, memory constraints, and network latency. It monitors agent utilization and resource consumption across your swarm to surface which layer—topology, caching, or concurrency—needs optimization.

After profiling detects bottlenecks, the Skill generates AI-powered optimization recommendations such as switching to hierarchical topology, enabling caching, or tuning concurrency. Recommendations can be applied automatically to improve swarm efficiency without manual rewiring.

Yes. Performance reports export in multiple formats suitable for automated ingestion and further analysis. Generate HTML reports for detailed visual review or machine-readable formats for integration into monitoring and analytics pipelines.

The Skill supports both real-time bottleneck detection and historical analysis. Use time-range and swarm-id options to analyze past performance issues or monitor current operations, enabling trend analysis and proactive optimization.

You need an active Claude Flow swarm and access to swarm metrics. The Skill requires Claude Flow as a dependency; ensure your swarm is instrumented to emit performance data for bottleneck detection and profiling to work.