Visualizes the chronological execution flow and performance metrics of multi-agent interactions within Claude Code.
The Trace skill provides deep observability into Claude Code sessions by generating detailed timelines and summaries of how hooks, keywords, agents, and tools interact. It helps developers debug complex multi-agent workflows, identify performance bottlenecks in specific tools, and understand the logic chains that lead to specific outputs. By surfacing mode transitions and event statistics, it transforms opaque AI processes into a transparent, audit-ready map of the execution flow, making it essential for managing multi-agent orchestration.
주요 기능
01Aggregate statistics on keyword detection and skill activation
02Visualization of agent mode transitions and flow patterns
03Performance bottleneck identification for slow tools
047,661 GitHub stars
05Filtered tracing to focus on specific event types or sessions
06Chronological event timeline for agents, tools, and hooks
사용 사례
01Auditing the AI's decision-making path during a coding session
02Profiling tool performance to reduce execution latency
