# Extract Clean Web Markdown

Web pages often include navigation, ads, and layout text that waste tokens. This skill uses Defuddle CLI to extract cleaner Markdown for Claude, Codex, and Claude Code.

## Install

```bash
npx skillstore add kepano/defuddle
```

## Metadata

- Status: approved
- Slug: kepano-defuddle
- Skillstore revision: r1
- Version status: missing
- Tree hash: c2be47e73771f5f8cc46cf0dd167d8c5240314ee315a78152080bedbf15684b0
- Author: kepano
- GitHub username: kepano
- License: MIT
- Repository: https://github.com/kepano/obsidian-skills/tree/main/skills/defuddle
- Ref: 0519034dad657fb1f7706e0550e962beeda73fdf
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/kepano-defuddle
- Manifest: https://skillstore.pages.dev/api/skills/kepano-defuddle/manifest

## Capabilities

- Extracts readable Markdown from a supplied web page URL.
- Removes common page clutter such as navigation, ads, and layout content.
- Saves extracted Markdown to a file when requested.
- Reads specific metadata fields such as title, description, and domain.
- Documents output format choices for Markdown, JSON, HTML, and metadata.

## Use Cases

- Review technical documentation: Extract clean Markdown from online docs before asking an assistant to summarize APIs or migration steps.
- Collect article text for research: Convert web articles into cleaner reading material with fewer navigation and advertising tokens.
- Prepare source material for notes: Save readable Markdown from a page so it can be reviewed, cited, or added to a notes workflow.

## Prompt Templates

### Extract one page

```
Use Defuddle to extract clean Markdown from this URL, then give me the main points.
```

### Save source material

```
Extract this page as Markdown with Defuddle and save it to a clearly named file in the workspace.
```

### Read page metadata

```
Use Defuddle to get the title, description, and domain for this page before summarizing it.
```

### Compare multiple sources

```
Extract clean Markdown from these URLs, then compare the claims and note any conflicts between sources.
```

## Limitations

- Requires the Defuddle CLI to be installed before use.
- Depends on the target page being reachable from the local environment.
- May not handle authenticated, scripted, or blocked pages correctly.
- Does not verify the accuracy of extracted page content.

## Best Practices

- Confirm the URL and command before running Defuddle on user-provided input.
- Prefer Markdown output for summaries, analysis, and documentation workflows.
- Use explicit file names when saving extracted content to the workspace.

## Anti Patterns

- Do not use it for pages that require private authentication without user approval.
- Do not assume extracted content is complete when the page depends on heavy scripting.
- Do not save output over existing files without checking the target path.

## Security Audit

- Audited at: 2026-07-09T11:27:25.08\+00:00
- Summary: The skill is a short Defuddle CLI usage guide with no prompt injection text or hidden behavioral instructions found. Several static hits are Markdown formatting false positives, but the npm install command and Defuddle parse examples are real external command use.

## Stats

- Views: 208
- Downloads: 23
- Favorites: 0
- Popularity score: 0