web-reading

What problem does it solve?

This skill eliminates the massive time and token waste caused by flailing when scraping heavy JavaScript single-page applications (SPAs) and virtualized feeds, where naive DOM scraping returns empty results, pulls 400K+ character blobs, or scroll-churning destroys data, turning a 30-60 second read into a 10-15 minute token-heavy failure.

Core Features & Use Cases

• Prioritized extraction strategies: Uses the first working method from a ranked list (markdown extract tools, accessibility snapshot, compact JavaScript evaluation) to avoid wasted attempts on obfuscated SPAs.
• Virtualized feed handling: Grabs all currently rendered content in one pass to avoid scroll-churn that unmounts older items, with built-in guardrails to stop chasing arbitrary item counts.
• Compact output: Extracts only needed fields at the source to avoid bloated token usage, with anti-thrashing rules to stop after 2 consecutive failed extraction attempts.
• Cross-platform support: Works for static pages (via web_fetch markdown conversion), public JS pages (via Jina Reader routing), and logged-in heavy SPAs (via browser MCP + built-in page-to-markdown script). For example, if you need to pull the latest public posts from an X/Twitter feed, this skill will return clean, compact content in under a minute without wasting resources on failed selector attempts.

Quick Start

Use the web-reading skill to extract the latest 3 public posts from the X/Twitter profile for Neko Core.