Zum Inhalt springen
Kitbay
Einreichen
Skill

bigquery-pipeline-audit

MITAktualisiert am 11. Nov. 202641.032GitHub-Sterne

A review prompt packaged as a skill in GitHub's awesome-copilot collection.

Was der Skill vermittelt

A review prompt packaged as a skill in GitHub's awesome-copilot collection. It has the agent audit a Python pipeline that uses BigQuery and report in six parts: where billed jobs are triggered and whether a bytes-billed limit is set, whether dry-run and execute modes exist, how backfills loop over dates, whether queries prune partitions and avoid SELECT star and exploding joins, whether writes are idempotent through MERGE or staging tables, and whether failures and job statistics are logged. It ends with a pass or fail verdict, a patch list ordered by risk and worst-case cost estimates. It reads code only and does not connect to BigQuery.

Funktionen

  • Audit BigQuery job triggers for runaway cost
  • Check backfill loops and partition pruning
  • Review write idempotency and deduplication keys
  • Produce a pass or fail report with a patch list