[OSS Review] Defuddle

This article shares my research findings on an OSS library used in Giselle. While originally written as an internal memo, I've structured it to be helpful for other developers considering similar OSS adoption decisions.

1. Overview of Defuddle

• Name: Defuddle

• Summary: Defuddle is a JavaScript/TypeScript library that extracts the main content from web pages and returns a cleaned, standardized document. It removes clutter like navigation, headers, and comments to provide consistent HTML or Markdown output, serving as a modern, metadata-aware alternative to Mozilla Readability.

• Core Features:

  • Extracts main article content and cleans up surrounding clutter (ads, headers, footers).

  • Standardizes common elements like headings, code blocks, footnotes, and math into a consistent format.

  • Extracts rich metadata including title, author, publish date, hero image, and schema.org data.

  • Supports both browser and Node.js environments and can convert the final output to Markdown.

2. About the Organization: kepano

• GitHub Org / Maintainers: kepano (Steph Ango)

• Summary: Defuddle is maintained by Steph Ango (@kepano), an active open-source developer known for creating the popular "Minimal" theme and several plugins within the Obsidian ecosystem. His work focuses on tools for thought, content extraction, and user interface design.

• Development Activity: The project is actively maintained with frequent releases (latest is 0.6.6 as of August 2025). It has a healthy number of stars (~2.8k), a growing contributor base, and is a core dependency of the official Obsidian Web Clipper.

3. Software License: MIT

• License: MIT

• Notes: The MIT License is permissive, allowing for commercial use, modification, and distribution with very few restrictions. It is generally compatible with both open-source and proprietary projects, requiring only that the original copyright and license notice be included in copies of the software.

4. Key Use Cases & Examples

• Major Use Cases:

  • Obsidian Web Clipper: Defuddle is the core extraction engine for the official Obsidian Web Clipper, used to capture and clean web articles for personal knowledge management.

  • defuddle-cli: A companion command-line tool enables using Defuddle in batch processing, data pipelines, and server-side automation without writing application code.

• Potential Applications:

  • Building "read-it-later" services or "reader mode" features in applications.

  • Pre-processing web content for ingestion into RAG (Retrieval-Augmented Generation) pipelines, summarization tools, or search indexes.

  • Archiving web pages in a clean, standardized, and durable format.

• Example Usage in Giselle:

  • Add WebPage node by toyamarinyon · Pull Request #1027 · giselles-ai/giselle

Reference

• GitHub: https://github.com/kepano/defuddle

• OSSInsight: https://ossinsight.io/analyze/kepano/defuddle

• Obsidian Web Clipper: https://help.obsidian.md/web-clipper/troubleshoot

• Defuddle CLI: https://github.com/kepano/defuddle-cli

This article was written with the help of giselles.ai.