.claude/skills/validate/SKILL.md
Run QA validation checks on the recommendation pipeline output. Use after pipeline runs to verify data quality.
npx skillsauth add praveenmaiya/holley-rec validateInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Runs comprehensive QA checks on pipeline output.
bq query --use_legacy_sql=false < sql/validation/qa_checks.sql
Expected passing results: | Check | Expected | Status | |-------|----------|--------| | User count | ~450,000 | ✓/✗ | | Duplicates | 0 | ✓/✗ | | Refurbished items | 0 | ✓/✗ | | Service SKUs | 0 | ✓/✗ | | Min price | ≥ $50 | ✓/✗ | | Max price | ≤ $10,000 | ✓/✗ | | HTTPS images | 100% | ✓/✗ | | Score ordering | Valid | ✓/✗ | | Diversity (max 2/PartType) | Valid | ✓/✗ |
If any check fails:
# Find duplicate SKUs
bq query --use_legacy_sql=false "
SELECT email_lower, rec_part_1, rec_part_2, rec_part_3, rec_part_4
FROM \`auxia-reporting.temp_holley_v5_17.final_vehicle_recommendations\`
WHERE rec_part_1 IN (rec_part_2, rec_part_3, rec_part_4)
OR rec_part_2 IN (rec_part_3, rec_part_4)
OR rec_part_3 = rec_part_4
LIMIT 10
"
# Find low-price items
bq query --use_legacy_sql=false "
SELECT email_lower, rec_part_1, rec1_price
FROM \`auxia-reporting.temp_holley_v5_17.final_vehicle_recommendations\`
WHERE rec1_price < 50
LIMIT 10
"
# Check score ordering
bq query --use_legacy_sql=false "
SELECT email_lower, rec1_score, rec2_score, rec3_score, rec4_score
FROM \`auxia-reporting.temp_holley_v5_17.final_vehicle_recommendations\`
WHERE rec1_score < rec2_score
OR rec2_score < rec3_score
OR rec3_score < rec4_score
LIMIT 10
"
# Check HTTPS images
bq query --use_legacy_sql=false "
SELECT email_lower, rec1_image
FROM \`auxia-reporting.temp_holley_v5_17.final_vehicle_recommendations\`
WHERE rec1_image NOT LIKE 'https://%'
LIMIT 10
"
sql/validation/qa_checks.sql - Full validation suitedocs/pipeline_run_stats.md - Historical comparisontesting
Generate a team-facing weekly status update from STATUS_LOG.md and git history.
research
Compare Personalized vs Static treatment performance with unbiased methodology. Use for A/B analysis and treatment comparison.
testing
Show current pipeline and deployment status. Use for quick health check.
devops
Execute the v5.17 vehicle fitment recommendation pipeline. Use when user asks to run the pipeline, refresh recommendations, or generate new recs.