The CDGC Data Quality Score server lets AI agents retrieve data quality scores for business and technical assets using Data Governance and Catalog.
This MCP server returns the following types of data quality scores:
•Single, overall score. This is a derived value.
•Per-dimension score. This is a stored value.
•Per-dimension acceptance breakdown. The server returns rule counts with values of Good, Acceptable, and Not Acceptable.
Key capabilities
This MCP server offers the following key capabilities:
•Overall asset score. Returns a governed asset's single, overall data-quality score. This one derived value is computed as the mean of its per-dimension scores.
•Per-dimension scores. Reads each quality dimension's score as stored. For a parent asset, this is the aggregated roll-up across its children.
•Acceptance-band breakdown. Expands any dimension into its underlying rule counts by band. The acceptance values are Good, Acceptable, and Not Acceptable.
Use cases
Use this MCP server to address the following use cases:
•Asset quality assessment. Fetch an asset's per-dimension scores and bands to report which dimensions pass or fail before trusting the data downstream.
•Dimension health check. Read a governed asset's overall score, then expand the weakest dimension's acceptance-band breakdown to see how many underlying rules have acceptance values of Good, Acceptable, or Not Acceptable.
•Rule occurrence check. Score a single rule occurrence to report its score, total and failed row counts, dimension, and last-run refresh date.
Tools
The following table describes the tools available on this MCP server:
Tools
Description
get_asset_score
Retrieves the scores for a single business or technical asset, including the following asset types:
- Business Area
- Business Dataset
- Business Term
- Data Element
- Domain
- Geography
- Legal Entity
- Metric
- Policy
- Process
- Project
- Rule Template
- Subdomain
- System
- Technical Dataset
Use the CDGC Metadata Search MCP to resolve names.
Note:
In AI Agent Engineering, the MCP server tools are called "actions."