The API includes the following metrics:

• How many records were processed?

• How many had errors?

• Which operators were involved?

• Was this a full failure or only partial?

Filters

• Session ID (per run)

• Operator ID (which operator caused issues)

• Error code (what kind of failures you saw)

Usage

• Replace missing context in status notifications. When a session transitions to FAILED or COMPLETED, your system can immediately call the Summary API for:

  • outputCount vs errorCount: Is this a small spike or a catastrophic failure?

  • Breakdown by operatorId: Identify the operator that is causing the error.

  • Breakdown by errorCode: Analyze if this is a configuration issue, data quality issue, or something else.

• Automated decision-making:

  • If only 0.5% of records failed, log and proceed.

  • If >10% failed in a single operator, trigger a rollback, alert an on-call channel, or pause downstream processing.

  • If failures are concentrated in a single error code, send that error class to a remediation pipeline.

• Operational dashboards:

  • Feed the response into BI or observability tools.

  • Track trends over time: error rate per operator, run performance, and the sessions that are the most problematic.