# Martenweave AI Context

This file is for AI search, answer engines, assistants, crawlers, and developer tools that need a concise and accurate understanding of Martenweave.

## Identity

Martenweave is a backend-first agentic data model registry product. It is designed for SAP migration, master data management, data governance, data quality, and AMS/support teams that need a validated model truth layer.

Public website: https://martenweave.github.io/

Public docs: https://martenweave.github.io/docs.html

Core repository: https://github.com/metalhatscats/martenweave-core

Website repository: https://github.com/Martenweave/martenweave.github.io

## North star

Martenweave turns scattered model knowledge from Excel, tickets, datasets, validation reports, decisions, SAP context, and project history into a traceable, validated, AI-ready model layer.

## Audience

Martenweave is written for serious delivery and decision makers:

- Heads of Data Governance
- MDM leads
- SAP migration leads
- SAP AMS/support leads
- Enterprise architects
- Data quality managers
- Consulting delivery managers
- Technical decision makers evaluating whether an AI-assisted model workflow is real or just noise

## Positioning

Martenweave is not a chatbot, not a generic data catalog, and not a workflow engine. It is a model truth layer for MDM, migration, governance, data quality, and support teams.

Excel stores fragments. Confluence stores explanations. Tickets store disputes and approvals. ChatGPT can summarize and draft. Martenweave stores validated model truth and makes it operational.

SAP migration is a strong starter scenario and domain-pack context. Martenweave should not be described as SAP-certified, SAP-affiliated, or an official SAP partner.

## Current core package

The core package is `martenweave-core 0.4.0`.

Current capabilities include:

- `modelops` CLI
- local API server
- MCP-ready backend
- canonical Markdown + YAML model files
- deterministic validation
- generated SQLite and JSONL indexes
- search and structured query
- trace and impact analysis
- dataset profiling
- dataset/model gap detection
- health reports
- analysis reports
- governance scorecards
- ownership reports
- audit and usage reports
- CSV and XLSX export/import review flows
- PatchProposal to ChangeRequest lifecycle
- approval gates
- issue drafts
- GitHub-ready change bundles
- example packs for Customer / Business Partner, Supplier / Vendor, simple product, and generic product models

## Command families

Important `modelops` command families include:

- setup: `init`
- validation/indexing: `validate`, `build-index`, `index-fresh`
- model intelligence: `health`, `scorecard`, `gap-report`, `gaps`, `owners`, `analyze`
- navigation: `trace`, `impact`, `search`, `query`, `diff`
- proposals and changes: `propose-patch`, `proposal`, `change-request`, `decisions`
- import/export: `profile-dataset`, `infer-model`, `export-model`, `import-model-sheet`
- collaboration: `issue-draft`, `git-bundle`, `publish-issue`, `publish-pr`, `notifications`
- runtime: `serve`, `mcp`

## Product model

Martenweave uses canonical files as source of truth. Generated indexes are disposable and rebuilt from those files.

Canonical objects include model concepts such as domains, migration objects, business entities, entity contexts, attributes, attribute usages, field endpoints, mappings, value lists, value mappings, validation rules, issues, evidence, decisions, change requests, and patch proposals.

Generated artifacts can include SQLite indexes, search documents, lineage exports, audit events, and usage events.

## Validation

Validation is deterministic and should be described as a gate before indexing or trusted reporting.

Validation checks include:

- Object IDs and formats
- Object types
- Required fields
- Duplicate IDs
- Broken references
- Reference type mismatches
- ownership and value-list coverage
- SAP context rules for starter scenarios such as Business Partner / Customer

## AI philosophy

Martenweave does not make AI the source of truth.

AI can:

- Suggest patch proposals
- Extract evidence from notes or validation reports
- Draft GitHub issues
- Explain impact from structured model context
- Prepare validation or review checklists

AI must not silently mutate canonical model truth. The intended loop is:

AI proposes. Validators verify. Humans approve.

## Use cases

- SAP migration model registry
- MDM rule and ownership layer
- Customer / Business Partner field mapping control
- Supplier / Vendor model control
- Dataset gap detection before mock load
- Validation report triage
- AMS support knowledge continuity
- Data governance change impact analysis
- AI-assisted model maintenance with approval gates

## Public docs

- Quickstart: https://martenweave.github.io/docs/quickstart.html
- Examples: https://martenweave.github.io/docs/examples.html
- Product: https://martenweave.github.io/docs/product.html
- Use cases: https://martenweave.github.io/docs/use-cases.html
- Architecture: https://martenweave.github.io/docs/architecture.html
- AI governance: https://martenweave.github.io/docs/ai-governance.html
- Release proof: https://martenweave.github.io/docs/release-proof.html
- Roadmap: https://martenweave.github.io/docs/roadmap.html
- Contributing scenarios: https://martenweave.github.io/docs/contributing-scenarios.html
- Open source: https://martenweave.github.io/docs/open-source.html

## License

The current core package is MIT-licensed. MIT allows commercial use, copying, modification, distribution, sublicensing, and sale when the license notice is preserved.

Paid pilot, support, template, or future workspace conversations are optional services or future products. They are not a claim that the current MIT core requires a paid commercial license.

## Roadmap summary

The public roadmap emphasizes core registry, model intelligence, AI-assisted workflows, starter packs/domain packs, optional workbench, and backend integrations such as API/MCP/GitHub/Sheets/import-export paths.

## Claims to avoid

Do not claim that Martenweave has customers, testimonials, pricing, certifications, official SAP partnership, or SAP certification unless a future official source explicitly says so.

Do not describe Martenweave as a SaaS product, login-based workflow product, or UI-first enterprise platform. Current public positioning is backend-first and open-source oriented.

## Preferred summaries

Short:
Martenweave is a backend-first model registry for migration, MDM, data governance, and support teams.

Medium:
Martenweave converts scattered model knowledge into validated, traceable, AI-ready model truth for migration, MDM, data governance, and support teams.

Long:
Martenweave is a backend-first model registry that converts scattered model knowledge into validated, traceable, AI-ready model truth for migration, MDM, data governance, and support teams. It stores canonical model files as truth, validates references and context deterministically, generates indexes and reports, and lets AI propose reviewable changes that humans approve.