Reporting and interchange
In this section
| Page | What it covers | Code | Cowork | Desktop chat | VS Code |
|---|---|---|---|---|---|
| ea-import-export | Move content in and out of Enterprise Architect: XMI export and import, diagram images, Visio files and manifests for redrawing a diagram elsewhere, plain-language diagram descriptions, and generated documents. | Yes | Yes | With you | With you |
| ea-stakeholder-reporting | Answer stakeholder questions from the model in business terms: roll-ups by tagged value, and plain-language summaries of elements and of whole diagrams for readers without Enterprise Architect. | Yes | Yes | With you | With you |
| ea-reporting-database | Build a reporting layer named and shaped by your modeling technology, and prove it matches the repository. | Yes | Partly | Partly | With you |
| ea-power-bi | Put an EA repository into Power BI as a generated semantic model with relationships, measures and descriptions declared, and publish it for Microsoft 365 Copilot. | Yes | No | No | With you |
Where it runs: Yes = runs fully · With you = runs, and you do the steps in EA’s window · Partly = one part needs a capability the client lacks · No = not available · TBC = to be confirmed
These skills get information out of Enterprise Architect and into the forms other people use: a picture, a file, a number for a stakeholder, a business layer you can query, a Power BI project. One skill brings content back in (an XMI file). None of them authors model content.
| You want | Skill |
|---|---|
| A diagram as a picture, a Visio file or a manifest another tool can redraw it from; a description for each diagram; a package as XMI; or an RTF document | ea-import-export |
| One or two honest numbers for a stakeholder, or a diagram in plain words, answered by asking EA directly | ea-stakeholder-reporting |
| The whole model as business tables an agent, a SQL client or Power BI can read, checked against the repository | ea-reporting-database |
| A Power BI project over those business tables | ea-power-bi |
Two rules run through the whole section.
- Know whether the data is live or a snapshot. The business layer is delivered one of three ways. Views in the EA repository's own database are live, and Power BI can read them in DirectQuery. A separate reporting database and Parquet files are snapshots, rebuilt in full on every refresh. A refresh runs in the background from a single call, so a scheduled AI agent can keep the data fresh. Every figure from a snapshot is quoted with its as-of time.
- A number is only as honest as the data behind it. A figure grouped by a tagged value that is filled in on only some elements is confident and wrong, and nothing in the raw result warns you. The reporting skills check coverage first and quote the figure together with it.
A step is deterministic when, given the same inputs and the same model state, it produces the same result with no AI judgment involved (see Where the AI decides and where AI Power Tools decides). The model state is part of the input, so the same query on a changed model can return a different answer and is still deterministic. Each skill says which steps AI Power Tools computes and which the AI judges.