Dialectic Layer

↓

Drop the canonical model here

or choose a file from this machine

.yaml.json .md.xlsx .csv.dbml .drawio

No model loaded

Paste it instead

Fetch from a URL

A raw file over HTTPS. Re-fetched on a schedule if the model is maintained elsewhere, so a revision here follows a revision there.

Pull from a repository

Repository

Path

Branch

On push

Where the model is already under version control, this is the route that keeps it there. A push opens a revision here rather than replacing the live one.

What comes out

Whatever goes in, this is what the platform stores. YAML, because the descriptions are prose: unreadable in JSON, unmanageable in a spreadsheet cell.

model: ict-assurance revision: 2026-07-15 use_case: DORA control assurance entities: - name: ict_asset_inventory definition: A single ICT asset owned or used by the entity aliases: [ICT_Asset_Register, CMDB_Asset_Record] fields: - name: asset_id type: string required: true description: Unique identifier for the ICT asset; used for dependency mapping and cross-referencing in risk assessments example: ASSET-SRV-2024-0012 - name: criticality_classification type: enum values: [CRITICAL, HIGH, MEDIUM, LOW] required: true description: Business criticality rating; determines inclusion in network severance capability verification example: CRITICAL relationships: - from: access_review_record.asset_id to: ict_asset_inventory cardinality: many-to-one meaning: access is granted to an asset

description is the field that matters most later: it is what the matcher reads when it decides whether a customer’s KRITIKALITAET column is this field or some other one.

Validation

Nothing is stored until it passes

What would be created

Entities 66ok

Fields 936, of which 673 requiredok

Relationships 60 resolved to a target entityok

Aliases 234 alternative names across 45 entitiesok

What is wrong with it

None of these stops the load. They are recorded against the revision so the same list is on screen afterwards, rather than being discovered six weeks later by whoever tries to use the model.

26 entities no definition Shape only. The matcher has a name and fields to work with and no sentence saying what the thing is.

46 references no target The field says it points at something and does not say at what. Cannot be drawn, cannot be traversed.

19 entities unreachable No relationship in or out. No question spanning one of these and anything else can be answered.

3 clusters not connected The model is in three pieces of 39, 4 and 4. Nothing joins them.

Loading records the four findings above against this revision.

Revision	Came from	Entities	Fields	Change
2026-07-15	Markdown dictionary, dropped	66	936	136 duplicate labels consolidated into existing entities	Not loaded
2026-04-02	Excel workbook, dropped	62	874	First load, from the delivery contract	Superseded

What comes out

Validation

Revisions