audit-schema-gaps

LinkML validation passes incorrect YAML records when run in open mode or when exit codes are swallowed, allowing silent schema validation drift. Closed-schema enforcement with strict instance validation is required to detect unexpected fields and ensure records cannot hide failures.

Run a schema audit using programmatic probes to check identifier policies, naming divergence across ontology fields, requiredness inconsistencies, orphan enums, and broken range references. This detects structural schema gaps that allow invalid data to pass validation undetected.

Scan pipeline modules that write YAML to verify dry-run support and in-pipeline validation before writing records. A pipeline assurance audit checks whether safeguards are wired into existing just targets to prevent unsafe writer behavior.

Yes, running a strict schema audit requires uv, linkml-validate, pyyaml, and just. These dependencies enable closed-mode LinkML validation, YAML parsing, and execution of the audit targets needed to detect schema drift.

Execute an end-to-end schema and records audit to produce machine-readable TSV outputs and narrative reports. This generates a prioritized fix backlog starting with the CI gate, categorizing instance failures, schema gaps, and pipeline writer issues.

Without closed-schema enforcement, LinkML validation operates in open mode and silently accepts broken YAML records with unexpected fields. This limitation allows schema drift and data quality issues to go unnoticed because validation exit codes are effectively swallowed.