Import Sources Guide

Unsterwerx can import data from multiple external sources, each with a dedicated adapter that normalizes the data into the standard document format.

Available Adapters

Source	Type	Input Format	Default Weight
Local Filesystem	`local`	Directory tree	2
ChatGPT	`chatgpt`	ZIP export with `conversations.json`	1
Notion	`notion`	ZIP workspace export with markdown	2
Obsidian	`obsidian`	Vault directory with wikilinks	2
Telegram	`telegram`	Export directory with `result.json`	1

The local adapter scans a directory tree for supported file types (PDF, DOCX, XLSX, PPTX, TXT, CSV, Markdown, SQL).

bash

unsterwerx import run /path/to/documents

This is equivalent to unsterwerx ingest /path/to/documents. The ingest command uses the local adapter internally.

Export your ChatGPT data from chat.openai.com → Settings → Export data. You'll receive a ZIP file containing conversations.json.

bash

unsterwerx import run ~/Downloads/chatgpt-export.zip --source-type chatgpt

Each conversation becomes a separate document in Unsterwerx with both user messages and assistant responses preserved as canonical content.

Export your Notion workspace: Workspace Settings → Export all workspace content → Markdown & CSV format.

bash

unsterwerx import run ~/Downloads/notion-export.zip --source-type notion

Each page becomes a separate document. Nested pages preserve their hierarchy in the document name.

Point the importer directly at your Obsidian vault directory:

bash

unsterwerx import run ~/Documents/MyVault --source-type obsidian

Wikilinks ([[Page Name]]) are resolved and markdown files are imported as documents.

Export your Telegram data via Desktop: Settings → Advanced → Export Telegram data. Choose JSON format.

bash

unsterwerx import run ~/Downloads/telegram-export --source-type telegram

The adapter reads result.json and imports messages as documents.

Each source type has a default trust weight reflecting the typical reliability of that source:

Academic (weight 5): Peer-reviewed papers, dissertations
Government (weight 3): Official publications, regulations
Curated / Local (weight 2): Manually organized document collections
AI-generated / ChatGPT (weight 1): AI-generated content, conversation exports

Override the weight for a specific import:

bash

unsterwerx import run ~/Documents/verified-papers --weight 5

View registered sources and item counts:

bash

unsterwerx import status

View import history:

bash

unsterwerx import history

Preview what would be imported without modifying the database:

bash

unsterwerx import run --dry-run /path/to/source