Tools Reference

All tools are read-only. They retrieve data from Datadog but never create, update, or delete anything.

Monitors

List, search, and inspect Datadog monitors.

Tool	Parameters	Description
list_monitors	query?: string, limit?: number (default: 25, max: 100)	Lists monitors, optionally filtered by query string (e.g., type:metric status:alert tag:service:web). Returns name, status, query, tags, and thresholds.
get_monitor	monitor_id: number	Retrieves full details for a specific monitor including status, query, thresholds, tags, and notification targets.
search_monitors	query: string, limit?: number (default: 25, max: 100)	Searches monitors using Datadog monitor search syntax. Example queries: type:metric status:alert, tag:service:web, notification:@pagerduty.

Example prompts:

• "Are there any monitors currently in alert status?"
• "Find all monitors tagged with service:web."
• "Get the details and thresholds of monitor 12345."
• "Search for monitors that notify PagerDuty."
• "List all metric monitors in our production environment."

---

Metrics

Query timeseries metric data.

Tool	Parameters	Description
query_metrics	query: string, from: number, to: number	Queries timeseries metric data using Datadog's metric query syntax. Returns data points and summary statistics. Times are UNIX epoch seconds.

The query parameter uses Datadog's native metric query language. Examples:

• avg:system.cpu.user{host:web-01} — average CPU usage for a specific host
• sum:trace.servlet.request.hits{service:web-app}.as_count() — request count for a service
• max:system.mem.used{env:prod} by {host} — memory usage grouped by host

Example prompts:

• "Show me the CPU usage for host web-01 over the last hour."
• "Query the average request latency for our API service over the past 4 hours."
• "What was the peak memory usage across production hosts today?"

---

Events

Browse infrastructure and application events.

Tool	Parameters	Description
list_events	start: number, end: number, limit?: number (default: 25, max: 100)	Lists events within a time range. Events include deployments, alerts, comments, and other activity. Times are UNIX epoch seconds.
get_event	event_id: number	Retrieves a specific event including title, text, tags, and related resources.

Example prompts:

• "What events happened in our infrastructure in the last 24 hours?"
• "Show me recent deployment events."
• "Get the details of event 67890."