veracity-audit

You can audit a data pipeline for source reliability and scoring math integrity by dispatching parallel specialized sub-agents that validate ingestion error handling, scoring formula boundaries, and declared quality gate enforcement, producing evidence-backed findings without manual code review.

Data pipeline quality gate compliance verifies that declared quality checks match enforced checks, catching threshold mismatches and missing test coverage. You need to verify it in production workflows when teams must ensure data trustworthiness and prevent silent data loss from unhandled errors like HTTP 429s.

Validating scoring formula correctness and boundary conditions involves checking formula implementations against expected math logic and identifying out-of-range value risks. This skill applies dedicated data-scientist sub-agents with shared context to analyze scoring source directories and detect boundary or implementation errors.

Yes, checking for silent data loss and ingestion error handling across pipeline sources is a core audit function. The skill analyzes your pipeline source directory to identify unhandled ingestion errors, schema versioning issues, and fallback logic gaps across all declared data sources.

Running a full data pipeline audit requires explicit path flags for your scoring config, quality gate config, pipeline source directory, and scoring source directory. These paths allow the sub-agents to locate and validate the correct configuration and source code files end-to-end.

Automating data compliance audits for multi-step pipelines requires complete and accurate config paths for scoring and quality gates. The audit is limited to analyzing declared pipeline source and scoring directories, meaning undeclared data sources or unconfigured quality gates may not be detected by the sub-agents.