LiteBrowse is a specialized low-token webpage research skill designed to extract only the most relevant content from long or cluttered pages.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install litebrowse
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install litebrowse using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is LiteBrowse Skill?
The LiteBrowse skill is a precision-engineered utility for Openclaw Skills designed to solve the problem of high token consumption and noise during web scraping. Instead of dumping entire HTML documents into an LLM context, it utilizes a local relevance extractor to identify and isolate specific content blocks that match a user's query.
By prioritizing relevance over volume, LiteBrowse ensures that AI agents operate with higher accuracy and significantly lower costs. It is particularly effective for large documentation sites, enterprise pricing pages, and cluttered career portals where the essential information is often buried beneath navigation menus and footer content.
LiteBrowse Skill Use Cases
- Extracting specific pricing tiers and API limits from complex commercial websites.
- Identifying job requirements and location details from expansive career pages.
- Conducting deep-dive research into local HTML files or cached pages without loading irrelevant metadata.
- Minimizing token expenditure when an AI agent needs to perform multiple web lookups in a single session.
How LiteBrowse Skill Works
- The agent identifies the information target and converts it into a concise search query.
- The local web_relevance_extract.py script is executed, targeting either a live URL or a local HTML file.
- The extractor parses the source and retrieves the top-k most relevant blocks within a strict character budget.
- The AI agent processes only this compact, high-signal context to generate its response.
- If the initial data is insufficient, the agent performs a controlled expansion by refining the query or slightly increasing the character budget.
LiteBrowse Skill Setup
To integrate LiteBrowse into your environment for Openclaw Skills, ensure you have the extraction scripts located in your local directory:
# Verify script location
ls ./scripts/web_relevance_extract.py
# Run a test extraction
python3 ./scripts/web_relevance_extract.py "https://example.com" "target information" --max-chars 1200 --top-k 3 --format json
LiteBrowse Skill Data Schema & Taxonomy
LiteBrowse utilizes a structured extraction format to maintain data integrity. The primary metadata and content organization include:
| Parameter | Type | Description |
|---|---|---|
| source | String | The URL or local path to the HTML source file. |
| query | String | The search terms used to rank block relevance. |
| top-k | Integer | The number of distinct content blocks to return. |
| max-chars | Integer | The hard limit on the total character count returned. |
| format | String | Output format, typically 'json' for structured data or 'text' for raw blocks. |
LiteBrowse Skill Advanced Features
- Local Extraction Logic: Processes data locally to ensure only relevant snippets are sent to the LLM.
- Controlled Expansion: Allows for iterative budget increases (top-k or max-chars) to find missing data without full-page dumps.
- HTML File Support: Capable of analyzing pre-downloaded or locally generated HTML files to bypass network restrictions.
- Source Citation: Built-in discipline for citing specific blocks to distinguish between extracted facts and model inference.
SKILL.md
Loading
Related Openclaw Skills
cuihua-error-handler
An AI-powered assistant designed to automate error handling, recovery strategies, and fault tolerance in production systems.
Product Research (LLM Agent Edition)
An advanced AI-driven market analysis tool that leverages Sorftime MCP and LLM agents to perform multi-dimensional Amazon product research and decision evaluation.
MOVA Connector Setup
A powerful integration tool that replaces MOVA sandbox mocks with live production data from your own business systems.
Bidding Project Analysis Skill
An intelligent automation tool for extracting bidding data from Excel, managing it via MySQL, and generating insightful project reports.
Context Stabilizer
A powerful tool for maintaining character, world-building, and plot consistency in long-form AI-generated content.
KimiClaw
KimiClaw enables developers to power their OpenClaw environment and Claude Code CLI using the free, Anthropic-compatible Kimi K2.5 coding model.
Featured*
