| name | datadog-metrics |
| description | Access Datadog metrics in the us5 site with environment-provided credentials and the helper script. Use when Codex needs to discover active metric names, query timeseries data, inspect metric tags, or answer monitoring questions from Datadog metrics. |
Datadog Metrics
Use this skill for Datadog metric work against us5.datadoghq.com. The helper script reads credentials from environment variables, so the common path is to export them once and run the script instead of hand-writing curl commands.
Quick Start
Run scripts/datadog_metrics.py from the repo root or by absolute path.
Set credentials first:
export DATADOG_SITE=us5.datadoghq.com
export DATADOG_API_KEY=...
export DATADOG_APP_KEY=...
Common commands:
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py list-active --from now-24h --prefix validator_api.
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py query 'avg:validator_api.current_slot{env:prod}' --from now-30m
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py tags validator_api.current_slot
Workflow
1. Discover the metric name first
If the user gives a rough description instead of an exact metric name, start with list-active.
- Use
--prefix to narrow broad namespaces.
- Use
--limit to keep output small.
- Use
--from now-24h or another explicit window so the result is easy to justify.
Example:
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py list-active --from now-24h --prefix transaction_monitor.
2. Query a timeseries
Use query once you know the metric expression.
- Pass a Datadog query string exactly as Datadog expects it, for example
avg:metric.name{tag:value}.
- Prefer explicit windows such as
--from now-15m --to now.
- Add
--show-points N when the user needs the most recent raw samples.
- Add
--output json when another tool or script will consume the result.
Supported time formats:
- Relative:
now, now-15m, now-2h, now-7d
- Absolute ISO-8601:
2026-03-27T09:00:00Z
- Unix epoch seconds or milliseconds
Example:
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py query \
'avg:validator_api.current_slot{env:prod}' \
--from now-1h \
--show-points 5
3. Inspect tags for a metric
Use tags when the user needs to know which tags are available for slicing a metric.
- Start without filters.
- Add
--configured-only to show only configured tags when the full set is noisy.
Example:
python3 codex/skills/datadog-metrics/scripts/datadog_metrics.py tags validator_api.current_slot --configured-only
Operational Notes
- The script reads defaults from
DATADOG_SITE, DATADOG_API_KEY, and DATADOG_APP_KEY.
- Use
DD_SITE, DD_API_KEY, or DD_APP_KEY for one-off overrides, or pass CLI flags directly.
- Keep answers grounded in the queried window. If the user says "today" or "now," echo the actual time window you used.
- This skill is for metrics. If the task is really about logs, traces, dashboards, or monitors, use a different workflow.
