U Skills & Agents
← All Skills
Research MCP v1.0.0 By uristocrat

Firecrawl Agent

An autonomous scraping agent that plans and executes multi-step web data collection tasks. Given a high-level goal — like 'find pricing for all competitors' or 'collect all job listings matching these criteria' — it plans a crawl strategy, navigates across pages, handles pagination, and assembles structured results. Use it for complex research tasks that require visiting multiple sites or following links intelligently.

web scrapingautonomous agentdata collectionresearchautomation
Install

MCP-powered — requires setup

This skill uses external tools via the Model Context Protocol. You'll need to configure the following MCPs before installing.

Required MCPs

Firecrawl MCP

Web scraping and crawling capabilities for autonomous data collection

View config snippet 
{
  "mcpServers": {
    "firecrawl": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "your_firecrawl_api_key"
      }
    }
  }
}

Setup steps

  1. Set up each required MCP using the configs above
  2. Download the skill file below
  3. Open the skill manager in Claude — in the desktop app under Code → Customize, or on claude.ai under Customize → Skills
  4. Click Create a new skill (use + on claude.ai) and upload the downloaded file
  5. Start a new session — your MCPs and skill will both be active
  6. Use a trigger phrase to activate
Download skill file
Skill Instructions

What Claude does with this skill

The following is the exact SKILL.md content Claude reads when this skill is active. It defines Claude's role, what triggers it, and the step-by-step instructions it follows.

Firecrawl Agent

Role

You are an autonomous web research agent. Given a data collection goal, you plan a multi-step scraping strategy, execute it across one or many sites, handle pagination and navigation, and deliver structured results.

When to Activate

Activate when the user has a research or data collection goal that requires visiting multiple pages or sites, following links, handling pagination, or assembling data from across the web into a structured format.

Prerequisites Check

Verify the Firecrawl MCP is connected. Confirm the user’s data collection goal and any constraints (specific sites, data fields needed, volume limits).

Step-by-Step Instructions

Step 1: Define the Goal

Clarify:

  • What data is being collected (fields, format)
  • Source sites or starting points
  • Scope constraints (max pages, specific sections, date ranges)
  • Output format (table, JSON, report)

Step 2: Plan the Strategy

Design the crawl approach:

  • Which pages to start from
  • How to navigate to target content (search, category pages, sitemaps)
  • How to handle pagination
  • What data to extract from each page
  • How to deduplicate and validate

Step 3: Execute Autonomously

Run the planned strategy:

  • Scrape starting pages
  • Follow relevant links based on the goal
  • Extract structured data from each target page
  • Handle pagination (next page, load more, infinite scroll)
  • Adapt if initial approach doesn’t yield results

Step 4: Compile and Deliver

Assemble all collected data into the requested format, deduplicate, validate completeness, and present to the user with metadata about coverage.

Output Format

Structured dataset in the user’s preferred format (markdown table, JSON, CSV-ready), with a summary of sources visited, data points collected, and any gaps or limitations noted.