troubleshoot

Diagnose unexpected AI agent behavior by analyzing direct debug logs in JSONL files. This approach investigates issues like slow responses or incorrect tool usage by deriving root causes from evidence within discovery, llm_request, and agent_response event logs.

Chat agent instruction or skill loading failures are diagnosed by analyzing direct debug logs. Investigating discovery event logs pinpoints exact reasons for these failures, such as name mismatches or configuration errors, based on concrete log evidence.

Find the root cause of AI agent request latency by investigating direct debug logs. Analyzing JSONL log files for discovery, tool_call, llm_request, and subagent event types reveals the exact bottlenecks causing slow responses.

You can investigate skipped chat agent instructions without guessing by deriving conclusions from direct debug logs. Analyzing JSONL log evidence focuses on specific event types like user_message and agent_response to explain the skipped behavior accurately.

The best way to analyze debug logs for subagent invocation issues is to investigate direct JSONL log files. By examining subagent event types within the logs, you can derive evidence-based conclusions to explain unexpected tool or subagent behavior.

To troubleshoot AI agent issues in JSONL debug logs, look for discovery, tool_call, llm_request, agent_response, user_message, and subagent event types. Analyzing these specific log entries helps derive evidence-based conclusions about unexpected behavior.