CLAUDE.md Template: Data Governance Projects

Overview

A CLAUDE.md file placed in your project root gives build-cli persistent context about the project — its scope, tech stack, systems involved, and conventions — so you do not have to re-explain the context at the start of every session. This article provides a ready-to-use CLAUDE.md template for projects in the data governance space, covering Collibra asset management, MDM matching rules (Reltio / MRHub), and RTiS terminology mapping.

Copy the template below, fill in the placeholders, and commit it to your project root as CLAUDE.md.

Background

Data governance projects often span multiple systems (Collibra, Reltio, MRHub, RTiS) and require knowledge of domain-specific conventions — asset type hierarchies, matching rule logic, terminology ontologies, and review workflows. Without persistent context, build-cli cannot make informed suggestions about which system to target, what conventions to follow, or which data steward to involve.

A well-filled CLAUDE.md for a data governance project reduces the need to re-establish this context in every prompt, and helps the AI avoid suggestions that are inappropriate for the governance domain (e.g., suggesting hardcoding values into asset descriptions, or skipping data steward review).

Main Content

Copy this template into CLAUDE.md at your project root and replace all [PLACEHOLDER] values:

## Project

**[PROJECT NAME]** — [One sentence: data governance configuration, Collibra asset management, MDM rules, or terminology work.]

## Scope

- Primary system: [Collibra / Reltio / MRHub / RTiS / combination]
- Domain: [Data Governance / MDM / Reference Data / Terminology]
- What this project manages: [asset catalogue entries / matching rules / reference data tables / terminology mappings]

## Collibra context (if applicable)

- Environment: [PROD / UAT / DEV]
- Asset types in scope: [Data Set / Column / Business Term / ...]
- Workflows involved: [Certification / Stewardship / ...]
- Import format: [CSV import / API / manual]

## MDM context (if applicable)

- System: [Reltio / MRHub]
- Entity types: [Customer / Product / Location / ...]
- Matching approach: [deterministic / probabilistic / hybrid]

## RTiS / Terminology context (if applicable)

- Ontologies in scope: [e.g., SNOMED, LOINC, internal Roche]
- Mapping direction: [local terms → standard / standard → local]

## Conventions

- All descriptions written for business audience (no technical jargon in asset descriptions)
- Review required from data steward before publishing any changes to production
- Terminology mappings must be validated against RTiS before use in a data contract

## Known constraints

- [Known limitations, pending decisions, or areas requiring special care]

Summary

Place CLAUDE.md in the project root to give build-cli persistent context for every session.
Fill in all sections that apply to your project; comment out or remove sections for systems not in scope.
The “Conventions” section is particularly important: it prevents the AI from producing output that violates your data governance standards (e.g., technical jargon in business-facing descriptions, skipping data steward sign-off).
Commit CLAUDE.md to version control so the team uses consistent context across sessions.
See the other CLAUDE.md example templates in this hub for dbt, ETL pipeline, and SQL analytics projects.