Add-on: Range Rules Validation (Policy Engine)

Use this skill to implement the validation stage:

• consumes extracted_clauses
• compares against known rules/policies (“Range”)
• emits structured validation results with explanations, severity, confidence, and review flags

This stage should not depend on top-k retrieval over the document as the main method.

Inputs

Collect:

• RANGE_RULE_SOURCE: db (recommended) or repo-file (YAML/JSON checked into repo).
• VALIDATION_RESULT_SET: compliant|non_compliant|uncertain.
• SEVERITY_LEVELS: default info|low|medium|high|critical.
• UNCERTAINTY_REVIEW_THRESHOLD: default 0.6.

Data Contracts