Smart Web Fetch for Openclaw

A specialized tool that replaces standard web fetching with a high-efficiency Markdown cleaning pipeline for AI agents.

leochens
v1.0.0
Mar 4, 2026
29
5.7k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install smart-web-fetch

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install smart-web-fetch 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 Smart Web Fetch?

Smart Web Fetch is a robust content retrieval tool designed to replace the default web_fetch functionality in AI workflows. It automatically processes target URLs through high-quality cleaning services like Jina Reader, markdown.new, and defuddle.md to provide agents with pure Markdown instead of cluttered HTML.

By utilizing this skill from the Openclaw Skills library, developers can significantly improve the quality of information provided to LLMs. The tool focuses on extracting the core text while stripping away navigation menus, advertisements, and scripts, leading to much higher accuracy during summarization and research tasks.

Smart Web Fetch Use Cases

  • Reducing token consumption by stripping non-essential HTML from web articles.
  • Implementing reliable web scraping for AI agents with multi-service redundancy.
  • Automating research tasks where clean, formatted Markdown is required for documentation.
  • Enhancing agent performance by providing context that is free of UI noise and tracking scripts.

How Smart Web Fetch Works

  1. The agent receives a URL and calls the Smart Web Fetch script.
  2. The system attempts to fetch the content via Jina Reader to generate high-quality Markdown.
  3. If the primary service fails, it automatically falls back to markdown.new or defuddle.md.
  4. If all cleaning services are unavailable, the skill retrieves the raw HTML as a final fallback.
  5. The cleaned content is returned to the agent, formatted specifically for LLM consumption.

Smart Web Fetch Setup

To integrate this skill into your environment, use the provided Python script. No API keys are required for the default cleaning services.

# Basic Markdown fetch
python3 scripts/fetch.py "https://example.com/article"

# Fetch with JSON metadata output
python3 scripts/fetch.py "https://example.com/article" --json

To ensure your agent prioritizes this Openclaw Skills implementation, update your openclaw.json to deny the built-in fetch tool:

{
  "agents": {
    "list": [
      {
        "id": "your-agent",
        "tools": {
          "deny": ["web_fetch"]
"        }
      }
    ]
  }
}

Smart Web Fetch Data Schema & Taxonomy

The skill provides a structured response format to ensure compatibility with complex agentic workflows.

Property Description
success Boolean flag indicating if the request was successful.
url The specific proxy or cleaning service URL utilized.
content The processed Markdown content of the target page.
source The specific service that successfully provided the data (e.g., jina).
error Contains error messages if all fallback attempts fail.

Smart Web Fetch Advanced Features

  • Four-level fallback strategy ensuring high availability for web content retrieval.
  • Token optimization that saves between 50% and 80% on context costs per request.
  • Native support for both plain text Markdown and structured JSON output formats.
  • Zero-cost operation using free, high-performance web cleaning APIs.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*