Datadog Observability

The datadog-observability skill enables AI agents to query, analyze, and visualize real-time infrastructure metrics directly within the Datadog platform. It provides immediate access to logs, traces, and performance data for troubleshooting production environments.

When to use it

• Investigating sudden latency spikes in a microservices architecture by querying distributed traces.
• Correlating application errors with specific infrastructure events like high CPU usage or disk I/O saturation.
• Generating automated incident reports by aggregating log patterns and metric thresholds from the last hour.
• Validating the impact of recent deployments by comparing current service health scores against baseline performance.

Key capabilities

• Real-time metric retrieval across hosts, containers, and processes.
• Log search and filtering with support for complex query syntax.
• Distributed tracing analysis to map request flows between services.
• Dashboard creation and visualization of custom metric combinations.
• Alert rule evaluation to identify threshold breaches automatically.

Example prompts

• "Show me the average response time for the payment service over the last 15 minutes and highlight any instances exceeding 200ms."
• "Search error logs from the web tier in the past hour and summarize the most frequent exception types."
• "Create a dashboard view comparing CPU utilization across all database nodes versus their current memory usage."

Tips & gotchas

Ensure your AI agent has read-only API access configured before attempting to query sensitive production data. Be mindful of Datadog's rate limits when running high-frequency queries to avoid triggering throttling mechanisms.