defuddle

Fetch readable text content from any URL using the defuddle.md service. Use this skill whenever you need to extract content from URLs that are difficult to fetch directly — such as x.com/twitter posts, pages behind JavaScript rendering, or sites that block standard fetchers. Also use it as a general fallback when WebFetch fails or returns unusable results. Trigger on: reading tweets/X posts, fetching article content from tricky domains, "grab the text from this link", or any URL where direct fetching is unreliable.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "defuddle" with this command: npx skills add plimeor/agent-skills/plimeor-agent-skills-defuddle

Defuddle

Fetch readable web content as Markdown via the defuddle.md service.

When to use

  • Hard-to-fetch domains: x.com, twitter.com, sites with heavy JS rendering, anti-scraping protections, or paywalls
  • WebFetch fallback: when WebFetch returns errors, empty content, or garbled output
  • General content extraction: whenever you need clean, readable text from a URL

How it works

defuddle.md is a public service that extracts the main content from a webpage and returns it as Markdown with YAML frontmatter.

curl -sL "defuddle.md/<url>"

The <url> is the target URL without the protocol (no https://).

Examples

Fetch an article:

curl -sL "defuddle.md/example.com/blog/some-post"

Fetch a tweet / X post:

curl -sL "defuddle.md/x.com/username/status/123456789"

Usage notes

  • Always use -sL flags (silent mode + follow redirects)
  • Strip https:// or http:// from the target URL before appending
  • The response is Markdown text — you can use it directly or extract specific sections
  • If the response is empty or an error page, the target site may not be supported; fall back to WebFetch or inform the user

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Automation

blog-writing

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

blog-feedback

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

nix-coding-protocol

No summary provided by upstream source.

Repository SourceNeeds Review