Web Browsing

web-browsing

Claude Code skills for rigorous web research — enforces full-page content reading, prevents shallow snippet-only analysis.

• 🔍 Quick discovery — find relevant pages, get snippets, discover URLs for any topic
• 📄 Full-page reading — fetch complete page content as markdown for thorough analysis
• 🚫 HARD-GATE enforcement — prevents treating search snippets as sufficient evidence
• 🎯 Depth-tiered skills — choose the right level of rigor for your task

What is this?

This is a set of Claude Code skills that enforce rigorous web research methodology. It provides two depth-tiered skills — one for quick snippet-based scanning, one for full-page content research — ensuring AI assistants never treat search snippets as authoritative evidence.

Designed for Claude Code sessions where web research quality matters. Uses Brave Search for discovery and Apify RAG Web Browser for full-page content extraction.

Skill Tiers

| Skill | Depth | When to Use | |-------|-------|-------------| | web-search | Shallow | Quick scan — find pages, get snippets, discover URLs | | web-research | Deep | Full-page content fetching and thorough analysis |

How It Works

Each skill enforces a HARD-GATE appropriate to its depth:

| Tier | HARD-GATE Rule | |------|---------------| | Search | Snippets are for orientation ONLY — no substantive conclusions | | Research | MUST fetch full page content — snippets alone are PROHIBITED |

This prevents the common failure mode where an AI assistant treats search snippets as sufficient evidence for substantive claims.

Tool Responsibility Matrix

| Tool | Role | Output | |------|------|--------| | brave-search | Discovery — web, news, video search | URL, title, snippet (NOT full content) | | apify/rag-web-browser | Full-page content fetching | Complete page as markdown |

Quick Start

1. Configure MCP Servers

Add both servers to your MCP configuration (.mcp.json or Claude Desktop config):

brave-search

{
  "mcpServers": {
    "brave-search": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/brave-search-mcp"],
      "env": {
        "BRAVE_API_KEY": ""
      }
    }
  }
}

Get an API key at: https://brave.com/search/api/

apify

{
  "mcpServers": {
    "apify": {
      "command": "npx",
      "args": ["-y", "@anthropic-ai/apify-mcp"],
      "env": {
        "APIFY_TOKEN": ""
      }
    }
  }
}

Get a token at: https://console.apify.com/account/integrations

2. Install Skills

Clone this repository and add the skills directory to your Claude Code project:

git clone https://github.com/yogsoth-ai/web-browsing.git

3. Use in Claude Code

Invoke the skills by name in your Claude Code session:

• /web-search — quick snippet-based discovery
• /web-research — full-page content analysis

Available Tools

brave-search (Discovery)

| Tool | Description | |------|-------------| | brave_web_search | Search the web — returns ranked results with title, URL, and snippet | | brave_news_search | Search recent news articles by topic | | brave_video_search | Search for video content |

apify/rag-web-browser (Full-Page Fetch)

| Tool | Description | |------|-------------| | rag-web-browser | Fetch any URL and return full page content as markdown |

Usage

web-search (quick scan)

Use when you need to:

• Quickly assess what information exists on a topic
• Find relevant URLs for deeper research
• Get a rough overview from snippets
• Discover sources before committing to full reads

web-research (deep analysis)

Use when you need to:

• Understand a topic thoroughly from web sources
• Read full documentation, articles, or technical pages
• Compare information across multiple sources
• Draw substantive conclusions backed by full content
• Conduct competitive analysis or market research

Project Structure

web-browsing/
├── skills/
│   ├── web-search/
│   │   └── SKILL.md          # Shallow — snippet-based orientation
│   └── web-research/
│       └── SKILL.md          # Deep — full-page content analysis
├── tests/
│   └── integration-prompt.md # Live integration test prompt
├── assets/
│   └── repo-info.txt
├── README.md
├── .gitignore
└── LICENSE

Links

• 🐙 GitHub repository
• 🔍 Brave Search API
• 🌐 Apify RAG Web Browser
• 🔧 Model Context Protocol

Version

v1.0.0 — Two depth-tiered skills (web-search, web-research)

📄 License

[Apache-2.0](LICENSE)

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: yogsoth-ai
• Source: yogsoth-ai/web-browsing
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.