Public docs

Capabilities

Martenweave Core is a local-first, backend-first model governance layer. It turns approved model knowledge into canonical Markdown and YAML files, then makes that knowledge inspectable through deterministic services and derived outputs.

What works today

• Canonical model files: domains, entities, attributes, contexts, field endpoints, mappings, value lists, issues, decisions, ChangeRequests, and PatchProposals.
• Deterministic validation: identity, schema, cross-object references, and supported SAP context checks run before indexing.
• Derived access layers: SQLite and JSONL indexes, search, structured query, trace, impact analysis, reports, and a local read-only viewer can be rebuilt from canonical files.
• Evidence and readiness work: profile CSV/XLSX data, detect dataset-to-model gaps, prepare readiness and review outputs, and retain evidence separately from canonical truth.
• Controlled changes: AI and import flows create reviewable PatchProposals. Validators verify; humans approve; approved changes are recorded through ChangeRequests and Git-oriented bundles.
• Local integration: the CLI, bound local API, MCP server, and Workbench are integration surfaces around the same Core services.

Practical use cases

SAP migration

Compare mappings and representative datasets with the approved model before mock-load testing. Trace a field through source, target, rules, and relationships; then assess impact before a change.

MDM and MDG delivery

Keep business definitions, field mappings, value logic, ownership, and decisions connected without pretending that a mapping workbook or an MDM platform is the sole source of implementation knowledge.

Data governance and AMS

Use stable model objects and linked evidence to make recurring incidents, local exceptions, ownership questions, and post-go-live changes easier to investigate and review.

Boundaries

Martenweave is not a hosted SaaS tenant platform, enterprise MDM replacement, generic chatbot, generic workflow engine, or direct SAP write-back tool. The local Workbench does not own model truth; canonical files remain authoritative.

Next step

Read the pilot-project approach, see how an engagement works, or contact the team with a representative model slice or migration question.