Investigation Tasks allow agents to automatically investigate issues in the background, just like an engineer. They focus on discovering the root cause of failures by analyzing resource states, cluster activity, dependencies, and alerts. Agents analyze resource states, cluster activity, and dependencies to identify the underlying cause of issues. The feature connects with observability tools including Prometheus, Alertmanager, Datadog, New Relic, GitHub, Docker, Trivy, and many more to enrich investigations with broader context. It surfaces clear, actionable insights that point engineers to the likely cause and next steps, not just raw signals. All tasks run in the background as read-only operations, providing you with a comprehensive task report in seconds.

​
How to Use

Investigation Task interface showing resource selection and analysis options

Investigation Task dialog for analyzing cluster issues

To start an investigation:
  1. Use the keyboard shortcut (Command + E on macOS, Ctrl + E on Windows)
  2. This opens the Investigation Task Dialog
  3. Pick the resource from resource context
  4. Select logs from the particular container
  5. Choose your preferred model
  6. Ask the agent to investigate
  7. The investigation runs in the background, performing read-only analysis
  8. Receive a detailed task report within seconds

​
Investigation Process

Detailed task report displaying analysis results and remediation steps

Task report showing investigation findings and recommendations

The agent performs the following steps automatically:
  1. Resource Analysis: Examines the selected Kubernetes resource configuration and current state
  2. Log Analysis: Reviews container logs for errors, warnings, and patterns
  3. Dependency Mapping: Identifies related resources and potential impact chains
  4. External Data: Queries configured monitoring and observability tools for additional context
  5. Root Cause Identification: Correlates findings to determine the most likely cause
  6. Remediation Suggestions: Provides specific next steps to resolve the issue