Metric issue tool output omits alert rule details (query, aggregate, condition, evaluated value), leaving the agent without the context it needs to help root-cause an alert.
The Sentry UI surfaces this clearly (see example: dataset, aggregate count(), query process_shard OR OutboxFlushError, interval, threshold, evaluated value), but none of it is included in the MCP tool response for metric_issue type issues.
- Issue type:
metric_issue / category: metric
- Missing fields: alert rule query, dataset, aggregate function, interval, threshold condition, evaluated value
- Impact: agent responses are effectively useless for metric alerts — no query context means no ability to investigate signal vs. noise, trend, or root cause
- The event output currently includes only generic tags (
level, interface_type) and trace info, with no alert rule payload
Metric issue tool output omits alert rule details (query, aggregate, condition, evaluated value), leaving the agent without the context it needs to help root-cause an alert.
The Sentry UI surfaces this clearly (see example: dataset, aggregate
count(), queryprocess_shard OR OutboxFlushError, interval, threshold, evaluated value), but none of it is included in the MCP tool response formetric_issuetype issues.metric_issue/ category:metriclevel,interface_type) and trace info, with no alert rule payload