How It Works

How It Works

PowerAutoDocs reads your Power Platform solution files and generates documentation from them. It works entirely from the files already in your Git repository — no connection to a live environment is needed at documentation time.

The two run modes

Running locally

You run PowerAutoDocs from your own machine using a single command. This is useful when you're setting things up for the first time, testing your config, or generating a one-off Word or PDF document.

You need Node.js installed on your machine. The output goes wherever you've configured it — a local folder for Word/PDF, or straight to your ADO Wiki.

→ Running Locally

Running via ADO Pipeline

You set up a pipeline in Azure DevOps that runs PowerAutoDocs automatically. Once it's in place, documentation updates every time the pipeline is triggered — no manual steps needed.

This is the recommended approach for ongoing client projects. The pipeline runs on ADO's own agents so nothing needs to be installed on anyone's machine.

→ Running via ADO Pipeline

What it reads

PowerAutoDocs reads from unpacked solution files — the XML and JSON files produced when you unpack a solution ZIP using the Power Platform CLI (pac solution unpack). These files are committed to your repo and are the source of truth for all generated documentation.

What it produces

Every run goes through the same pipeline:

Unpacked solution XML / JSON
        ↓
    Parsers — one per component type, extract structured data
        ↓
    IR — typed intermediate representation, the internal data model
        ↓
    Enrichment — ER diagrams, flow diagrams, optional AI summaries
        ↓
    Renderers — produce format-agnostic document content
        ↓
    MarkdownSerializer  →  ADO Wiki (published via REST API)
    DocxSerializer      →  Word .docx file
    PdfSerializer       →  PDF file

The three output formats are independent — you can enable any combination of them in your config.