search-extract

name	search-extract
description	Web search and structured data extraction. This skill should be used when the user asks to search the web, extract structured data from pages into JSON, research a topic using online sources, or perform competitive analysis.

Search & Data Extraction

This skill covers web search and structured data extraction with Firecrawl — searching with content, extracting data with JSON schemas, and LLM-powered extraction.

Available Tools

Tool	Description
search	Search the web and get results with content
extract_data	Extract structured data from URLs

Web Search

Basic Search

Tool: search
Input: {
  "query": "best REST API design patterns 2024",
  "limit": 5
}

Returns: Array of results with URL, title, description, and full page content.

Search with Geographic and Language Filters

Tool: search
Input: {
  "query": "best CRM software",
  "limit": 10,
  "lang": "en",
  "country": "us",
  "location": "San Francisco, CA"
}

Search with Time Filter

Tool: search
Input: {
  "query": "AI automation tools",
  "limit": 5,
  "tbs": "qdr:m"
}

Time filter values (tbs): qdr:h (past hour), qdr:d (past day), qdr:w (past week), qdr:m (past month), qdr:y (past year).

Search with Source Filtering

Tool: search
Input: {
  "query": "workflow automation comparison",
  "limit": 5,
  "sources": ["techcrunch.com", "producthunt.com"]
}

Search Parameters

Parameter	Type	Description
query	string	Search query (required)
limit	number	Max results (default: 5)
lang	string	Language code (e.g., "en", "de", "fr")
country	string	Country code (e.g., "us", "uk", "de")
location	string	Specific location (e.g., "New York")
tbs	string	Time-based filter (qdr:h, qdr:d, qdr:w, qdr:m, qdr:y)
filter	string	Additional search filter
sources	string[]	Limit to specific domains
scrapeOptions	object	Options for scraping results (formats, etc.)

Each result includes the full scraped content of the page, not just a snippet.

Structured Data Extraction

Extract Parameters

Parameter	Type	Description
urls	string[]	URLs to extract from (array)
prompt	string	What to extract (required)
systemPrompt	string	System-level extraction instructions
schema	object	JSON schema for structured output
allowExternalLinks	boolean	Follow external links during extraction
enableWebSearch	boolean	Augment extraction with web search
includeSubdomains	boolean	Include subdomain pages

Extract with JSON Schema

Tool: extract_data
Input: {
  "urls": ["https://store.example.com/product/123"],
  "prompt": "Extract product details",
  "schema": {
    "type": "object",
    "properties": {
      "name": { "type": "string", "description": "Product name" },
      "price": { "type": "number", "description": "Price in USD" },
      "description": { "type": "string", "description": "Product description" },
      "rating": { "type": "number", "description": "Average rating out of 5" },
      "in_stock": { "type": "boolean", "description": "Whether product is in stock" }
    },
    "required": ["name", "price"]
  }
}

Returns: { "name": "Widget Pro", "price": 29.99, "description": "...", "rating": 4.5, "in_stock": true }

Extract from Multiple URLs

Tool: extract_data
Input: {
  "urls": [
    "https://store.example.com/product/123",
    "https://store.example.com/product/456"
  ],
  "prompt": "Extract product name and price",
  "schema": {
    "type": "object",
    "properties": {
      "name": { "type": "string" },
      "price": { "type": "number" }
    }
  }
}

Extract with Web Search Augmentation

Tool: extract_data
Input: {
  "urls": ["https://example.com/company"],
  "prompt": "Extract company revenue and employee count",
  "enableWebSearch": true,
  "systemPrompt": "If data is not on the page, use web search to find it"
}

Extract List Data

Tool: extract_data
Input: {
  "urls": ["https://example.com/team"],
  "schema": {
    "type": "object",
    "properties": {
      "team_members": {
        "type": "array",
        "items": {
          "type": "object",
          "properties": {
            "name": { "type": "string" },
            "role": { "type": "string" },
            "bio": { "type": "string" }
          }
        }
      }
    }
  }
}

Extract with Prompt

Tool: extract_data
Input: {
  "urls": ["https://example.com/pricing"],
  "prompt": "Extract all pricing plans with their names, prices, and included features"
}

Common Workflows

Research a Topic

1. search(query, limit=10) -> Find relevant sources
2. Review results — identify most relevant URLs
3. scrape_url(best_url) -> Get full content for deep analysis

Extract Data from Multiple Pages

1. map_site(url) -> Discover product/listing pages
2. Filter URLs matching pattern (e.g., /products/*)
3. For each URL: extract_data(url, schema) -> Get structured data
4. Compile results

Competitive Analysis

1. search("competitor product features", limit=5) -> Find competitor pages
2. For each competitor: extract_data(url, features_schema) -> Get structured feature data
3. Compare features across competitors

Content Research

1. search(topic, limit=10) -> Get source pages
2. For top results: scrape_url(url) -> Get full content
3. Synthesize information from multiple sources

Schema Design Tips

1. Use descriptive field names — helps LLM extraction accuracy
2. Add descriptions to properties — guides the extraction model
3. Use required fields — ensure critical data is extracted
4. Keep schemas focused — extract one type of data per call
5. Use arrays for lists — team members, pricing plans, features
6. Test with one URL first — verify schema works before batch extraction

Best Practices

1. Search before scraping — search is faster for finding specific information
2. Use schemas for consistency — structured extraction is more reliable than free-form
3. Provide descriptions in schemas — helps the extraction model understand context
4. Start with simple schemas — add complexity gradually
5. Combine search + extract — search to find, extract to structure
6. Limit search results — 5-10 results is usually sufficient