skills.rest
somnia-chain
somnia-chainOFFICIAL
Explore all skills by this author →
Website𝕏@Somnia_NetworkGitHub

somnia-agents-llm-parse-website

Extract structured string or numeric fields from live webpages via markdown conversion.

2|2|Updated May 19, 2026
Search Tags:#structured data#web scraping#headless browser#llm extraction#on-chain agent#schema output#numeric parsing
One-click install
npx skills add https://github.com/somnia-chain/agentathon --skill somnia-agents-llm-parse-website-somnia-chain
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: somnia-agents-llm-parse-website
Source: https://github.com/somnia-chain/agentathon/tree/main/somnia-agents-skills/skills/somnia-agents-llm-parse-website
Command: npx skills add https://github.com/somnia-chain/agentathon --skill somnia-agents-llm-parse-website-somnia-chain

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It turns messy webpage content into structured, schema-shaped data without requiring a JSON API or manual copy-paste.

Core Features & Use Cases

  • Browser-based scraping to markdown: renders JavaScript-heavy pages in a real headless browser, converts them to markdown, then extracts fields from that content.
  • Structured extraction with typed outputs: supports string extraction and bounded numeric extraction (uint256) via ExtractString and ExtractANumber schemas.
  • Search or direct scrape modes: use resolveUrl=true to search across multiple results, or resolveUrl=false to scrape a stable direct URL for better consensus.

Quick Start

Extract the “best_drama” film title from the Golden Globes winners page by prompting ExtractString with the direct URL and resolveUrl set to false.

Frequently Asked Questions about somnia-agents-llm-parse-website

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract structured data from a webpage that doesn't have a JSON API?▼

Web scraping with a headless browser converts live HTML pages into markdown for on-chain LLM schema extraction. It handles JavaScript-heavy pages to pull structured string or bounded numeric fields without needing a JSON API.

What's the best way to parse award results and sports scores from live HTML pages?▼

Parsing award results and sports scores is done by providing a direct URL and schema-aligned keys to the scraper. It renders the page, converts it to markdown, and extracts the target fields using schema-aligned descriptions.

How do I scrape a website when I don't know the exact URL of the content?▼

Scraping unknown URLs requires setting resolveUrl to true to enable multi-page search. This mode searches across multiple results to locate the target content, though direct URL scraping with resolveUrl false offers better consensus.

Does web scraping work for extracting numeric values into a bounded schema?▼

Web scraping supports bounded numeric extraction via the ExtractANumber schema, outputting uint256 values. You can apply optional min-max constraints to ensure the extracted numeric data fits your required parameters.

Can I use LLM extraction to scrape dynamic JavaScript-heavy webpages?▼

LLM extraction works on dynamic JavaScript-heavy webpages by rendering them in a real headless browser. The rendered page is converted to markdown, allowing the LLM to reliably parse the structured fields.

What are the limitations of using search mode versus direct URL scraping for LLM extraction?▼

Search mode introduces extraction cost and search variance across multiple pages compared to direct URL scraping. Using resolveUrl false on a stable direct URL is recommended to avoid variance and achieve better consensus.

Related Skills

Explore More Data & Analytics Skills→
Largo2z9Largo2z9
1

define-specs

Create validated product specifications from URLs, documents, and operator inputs.

Community
Advanced
X-School-AcademyX-School-Academy
19

test-workflow-detect-number-string

Inspect a string and output numeric characters or the word "string".

Official
Basic
acnlabsacnlabs
45

economy

Record agent costs and income to compute financial health tiers.

Official
Advanced
kevinnftkevinnft
13

web-scraping

Extract structured data from static HTML, JS-rendered pages, and hybrid sites.

Community
Intermediate