data-refresh-eval
Installation
SKILL.md
Data Refresh & Eval Skill
Workflow for keeping the eval dataset fresh and running quality checks on agent responses.
Quick Start
cd ~/Code/skillrecordings/support/packages/cli
# Refresh dataset from Front (last 30 days, 200 responses max)
bun src/index.ts dataset build --since $(date -d "30 days ago" +%Y-%m-%d) --limit 200 --output data/eval-dataset.json
# Run routing eval
bun src/index.ts eval routing data/eval-dataset.json