List Quality Checker Agent
Flags missing required fields, low-confidence values, and structural inconsistencies in account lists.
What it is
A list that reaches the scoring model missing half its required fields produces unreliable scores, and the problem won't surface until a rep calls an account that was never a real prospect. This agent catches those issues before they propagate.
How it works
The agent runs a four-layer audit. Completeness check: every required field is present and non-null. Validity check: values conform to expected formats. Consistency check: cross-field logic holds, such as headcount band matching revenue range. Confidence check: enriched fields from external sources meet a minimum confidence threshold. Each failing record is annotated with the specific issues detected. Output includes an overall quality score, a row-level annotation file, and a summary report by field and issue type.
How it fits into the pipeline
Quality gate between the normalization and enrichment steps and the scoring and outreach layers. Lists that fail are returned for remediation. Only lists that pass are forwarded to the Firmographic Enricher and Intent Signal Aggregator.