AI-powered documentation for ETL pipelines — in seconds.

PipelineDoc reads your pipeline files and generates complete, professional Markdown documentation using AI. Supports Talend, Azure Data Factory (ADF), and Microsoft Fabric Data Pipelines.

Features

Architecture diagram — auto-generated Mermaid flowchart rendered directly in the preview
Full documentation — Purpose, Data Flow, Source & Target Systems, Transformations, Data Lineage, Impact Analysis, Risks, Operations
Multi-format support — Talend .item files, ADF pipeline JSON, Fabric pipeline JSON, ARM templates
Secure API key storage — keys stored encrypted via the system keychain (Windows Credential Manager / macOS Keychain), never in plain text
Timestamped output — every generation creates a new file, nothing is ever overwritten
Language support — English or Swedish output

Supported Pipeline Formats

Format	File type	How to use
Talend	`.item`	Right-click in Explorer
Azure Data Factory	`.json`	Right-click in Explorer
Microsoft Fabric	`.json`	Right-click in Explorer
ARM template	`.json`	Right-click in Explorer

Getting Started

1. Set your AI provider

Open the Command Palette (Ctrl+Shift+P) and run:

PipelineDoc: Set OpenAI API Key — enter your OpenAI API key
Then open Settings (Ctrl+,), search pipelineDoc, and set AI Provider to openai

Premium users: set the provider to premium and enter your license key — no personal API key required.

2. Generate documentation

Right-click any supported pipeline file in the Explorer panel and select:

PipelineDoc: Generate Documentation from .item file (Talend)
PipelineDoc: Generate Documentation from ADF / Fabric / Talend JSON

The rendered documentation opens automatically in Markdown Preview.

Output Example

Each generated document includes:

# MyPipeline

## Architecture Overview
[Mermaid flowchart]

## Purpose
## Data Flow
## Source Systems
## Target Systems
## Transformations
## Data Lineage
## Impact Analysis
## Risks
## Operations & Error Handling

Files are saved next to the source file with a timestamp: MyPipeline_doc_2026-06-21_14-30-00.md

Settings

Setting	Description	Default
`pipelineDoc.aiProvider`	`openai` / `anthropic` / `premium`	`premium`
`pipelineDoc.outputLanguage`	`english` / `svenska`	`english`

API keys are managed securely via the Command Palette — not stored in settings.

Requirements

VS Code 1.85 or later
An OpenAI or Anthropic API key (Freemium), or a PipelineDoc Premium license key

Privacy & Security

Pipeline data is sent to your chosen AI provider (OpenAI or Anthropic) for documentation generation
API keys are stored encrypted using VS Code's built-in Secret Storage (system keychain)
No data is stored by xHolst