stripfeed

What This Skill Does

The StripFeed skill acts as a powerful web content pre-processor for your AI agents. It intelligently fetches web pages, stripping away intrusive advertisements, navigation menus, boilerplate scripts, and layout noise to leave behind clean, structured Markdown. Because LLMs perform significantly better when provided with high-signal, low-noise input, this tool optimizes the quality of the retrieved information while simultaneously managing context windows. It provides granular feedback, including actual token counts for the extracted content, original page size, and the calculated token savings, allowing you to monitor your budget and context window usage in real-time.

Installation

To integrate this into your workflow, execute the following command within your terminal or OpenClaw environment: clawhub install openclaw/skills/skills/atlasopenclaw/stripfeed

Ensure that you have obtained your free API key from https://www.stripfeed.dev and set it as an environment variable in your system configuration: export STRIPFEED_API_KEY='your_api_key_here'. The skill uses this key to authenticate every request, ensuring you stay within your monthly usage limits.

Use Cases

Use StripFeed whenever your AI needs to process live web data. Common scenarios include: summarizing long-form news articles, extracting documentation from technical websites, monitoring specific blog feeds for updates, or aggregating content for research tasks. By using the 'selector' parameter, you can target specific content areas, such as the 'main' body or an 'article' tag, making it perfect for scraping complex sites while ignoring irrelevant sidebar noise.

Example Prompts

1. "Fetch the documentation from https://example.com/docs and summarize the primary configuration steps into a bulleted list."
2. "Get the latest news article from https://news-site.com/article-123. Ensure the output is under 2000 tokens and formatted as clean Markdown for my knowledge base."
3. "Refresh the data for https://tech-blog.com/post, bypass the cache to ensure I'm seeing the latest updates, and provide me with the token savings stats."

Tips & Limitations

To maximize efficiency, always default to the 'markdown' format unless you specifically require metadata, in which case use 'json' to access the full array of metrics including fetch latency and original token counts. Be aware that while the free tier is generous (200 requests/month), high-frequency scraping may require a Pro subscription. For dynamic websites that load content via JavaScript, ensure you understand the limitations of the scraper, as highly complex single-page apps may occasionally struggle to render content without specific CSS selectors.