consensys/ask-o11y-plugin/rendering-visualizations

Rendering PromQL queries as graphs

When providing PromQL queries, render them as interactive visualizations in the chat using this format:

your_promql_query_here

prometheus fences work identically.

Example:

sum by (service) (rate(http_requests_total{status=~"5.."}[5m]))
  / sum by (service) (rate(http_requests_total[5m]))

Choosing the visualization type (`viz` attribute) — match the data shape:

• timeseries (default) — any metric over time, rates, trends
• gauge — current value against min/max context (CPU %, memory utilization)
• stat — single aggregate values, counts, uptime
• table — multi-column rows, detailed label breakdowns
• piechart — parts of a whole, proportions
• barchart — category comparisons, rankings
• heatmap — distribution over time, histogram buckets
• histogram — value distribution in a single window

Rendering LogQL queries as log panels

When providing LogQL queries, render them as interactive log panels in the chat using this format:

your_logql_query_here

loki fences work identically.

Rendering TraceQL queries as trace panels

When providing TraceQL queries, render them as interactive trace panels in the chat using this format:

your_traceql_query_here

tempo fences work identically.

The title attribute is optional but recommended. The from and to attributes control the displayed time range (default: last 1 hour) and accept relative (now-24h) or absolute values.

Datasource selection — optional ds="<uid>" attribute on the opening fence line, using a UID from the datasource listing tools. Omit it to use Grafana's default datasource for that type. When several instances exist, either set the intended one as default in Connections or pass ds explicitly.