---
description: Evaluate coprocessor worker HPA scaling quality and generate report page
agent: scaling-evaluator
---

IMPORTANT: Prefer retrieval-led reasoning over pre-training-led reasoning.
For PromQL queries, scoring thresholds, API endpoints, and event detection:
ALWAYS use the methodology and API reference injected below.
NEVER guess from training data.

## Target

Evaluate COPROCESSOR WORKER HPA scaling for cluster $1
Vendor: $2 (default: aws)
Region: $3 (default: us-west-2)

## Required Output

This command MUST produce a **report HTML file** from the scaling evaluation.
The evaluation is NOT complete until all steps are done:

1. **Evaluate**: Query metrics, detect events, compute scores using the methodology below
2. **Create data file**: Write a JSON file with evaluation data (schema in report UI reference)
3. **Generate report**: Run `generate-report.sh <data-file> <template> <output>` to inject data into the HTML template
4. **Open**: View the generated HTML file in a browser

See the report UI reference for the data schema and template details.

## Shared Evaluation Methodology

@.opencode/skills/autoscaler-dashboard/SKILL.md

## Component: Coprocessor Worker

@.opencode/skills/autoscaler-dashboard/components/copr-worker.md

## Report UI Reference

@.opencode/skills/autoscaler-dashboard/references/report-ui.md

## O11Y Metrics API Reference

@.opencode/skills/o11y-metrics-api/SKILL.md