Guide · Agent skill
How to use the Defuddle Web Extractor agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The defuddle skill teaches an AI agent to fetch a web page and return only its main readable content as Markdown, with navigation, ads and other clutter removed. It is aimed at researchers, developers and anyone who wants an agent to summarise or archive an article. This guide explains what the skill does, how to install it, how to trigger it, and what it needs to run.
Defuddle Web Extractorkepano/obsidian-skills · MIT Download the skill Quick start
- 1Check Node.js and npm. Make sure Node.js and npm are installed, since the Defuddle CLI is installed with npm.
- 2Install the Defuddle CLI. Run npm install -g defuddle in a terminal to install the tool globally.
- 3Download and unzip the skill. Download the ZIP from this page and unzip it so the folder contains SKILL.md.
- 4Place the skill folder. Move the folder into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder, or upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
- 5Ask with a URL. Make a plain-language request that includes the web page address, such as asking for the article text as Markdown.
- 6Check the output. Compare the agent's result with defuddle parse <url> --md run in your terminal if you want to confirm what was extracted.
What is an agent skill and what does defuddle make the agent do?
An agent skill is a folder containing a SKILL.md file. The file gives an AI agent instructions for a specific task, and the agent loads them when a request matches the skill's description. This skill is named defuddle and is described as extracting clean Markdown from HTML pages with the Defuddle CLI.
When the skill is active, the agent uses the Defuddle command-line tool to fetch a page and strip out navigation, ads and other clutter, leaving the main readable text. The skill tells the agent to prefer this over its built-in WebFetch for standard web pages, because the authors say it reduces token usage. The skill can also pull out specific metadata such as the title, description or domain.
Install the defuddle skill
Download the ZIP file from this page and unzip it. The skill is a single file, SKILL.md, so the unzipped folder should contain that file.
For a personal install, place the unzipped skill folder in ~/.claude/skills/. To make it available only inside one project, place it in that project's .claude/skills/ folder.
Alternatively, upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
The skill does not install the Defuddle tool for you. Install the CLI separately as described in the next section.
- Personal: ~/.claude/skills/
- Project only: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Install the Defuddle CLI
The skill requires Node.js and npm. If the Defuddle CLI is not already installed, the skill's instructions say to install it globally with npm.
Run this command in a terminal:
npm install -g defuddleHow to trigger the skill with natural requests
You do not need a special command. Ask the agent in plain language to get, extract or save the content of a web page, and it can use the skill. Include the page URL in your request so the agent knows what to fetch.
Typical phrasings include asking for the article text from a link, a clean copy of a page saved as a file, or just the title or description of a page.
- Get the readable text of https://example.com/article as Markdown.
- Save this article to a Markdown file called content.md.
- What is the title and description of this page: https://example.com/post?
- Summarise this web page without the menus and ads: https://example.com/report
Commands the skill uses
The skill tells the agent to always use the --md flag for Markdown output. You can run the same commands yourself to check what the agent will get.
To save the output to a file, add -o with a file name. To extract a single metadata property, use -p with the property name. The output formats are Markdown (--md, the default choice), JSON with both HTML and Markdown (--json), HTML (no flag), and a specific metadata property (-p).
- Markdown: defuddle parse <url> --md
- Save to file: defuddle parse <url> --md -o content.md
- Title: defuddle parse <url> -p title
- Description: defuddle parse <url> -p description
- Domain: defuddle parse <url> -p domain
defuddle parse <url> --md
defuddle parse <url> --md -o content.md
defuddle parse <url> -p title
defuddle parse <url> -p description
defuddle parse <url> -p domainSafety notes: scripts and network access
This skill runs scripts and uses the network. The Defuddle CLI fetches the URL you give it, so the agent will make outbound requests to that address. Only ask the agent to process pages you are comfortable having fetched.
The skill only describes fetching public URLs. Its instructions do not cover logging in, so pages that need authentication or sit behind a paywall are outside what this skill covers. Review the skill files before installing if you need to confirm what runs on your machine.
- Runs scripts: yes
- Uses network: yes
- Only the single file SKILL.md is included
What people use it for
- Save the text of a long article as a clean Markdown note for a research library.
- Feed readable page content into a developer pipeline without navigation and ad markup.
- Ask an agent to summarise a report from a public web page without the surrounding clutter.
- Pull the title, description or domain of a page for a list or index.
Frequently asked questions
Is the defuddle skill free?
The skill is published under the MIT licence, as stated on this page. The Defuddle CLI is installed with npm, so check its own terms if you need specifics beyond what this guide covers.
What do I need to use the defuddle skill?
You need Node.js and npm to run npm install -g defuddle, which installs the Defuddle CLI globally. The skill also runs scripts and uses the network to fetch pages.
Where do I put the defuddle skill folder?
Place it in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder for one project. In the Claude apps you can instead upload the ZIP under Settings → Capabilities → Skills.
Can defuddle extract content from pages behind a login or paywall?
The skill only describes fetching the public URL and does not cover authentication. Pages that need a login or are behind a paywall are not covered by its instructions.
Why does the skill prefer defuddle over WebFetch?
The skill says Defuddle removes navigation, ads and other clutter from standard web pages, and the authors say this reduces token usage compared with WebFetch.
Will it work on single-page apps where content loads after the page renders?
The file gives no guidance for this case. Extraction depends on the HTML the tool receives, so content that loads only after rendering may not be captured.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Choose this skill if you want an agent to pull the readable text of public articles and pages into Markdown, for research notes, archiving or pipelines. Pick something else if you need to access logged-in or paywalled pages, or if your target content is built by JavaScript after the page loads, since the skill does not cover either case.
Download the skill