Web Research

Use this skill for quick web lookups, local searches, and page content extraction — faster and lighter than deep_research.

Available Tools

Tool	Purpose	Speed	Depth
web_search	General web search (Brave/Tavily/DuckDuckGo)	Fast (1–3s)	Light — top N results
brave_web_search	Alternative web search	Fast	Light
brave_local_search	Local businesses, places, "near me" queries	Fast	Light
run_crawl4ai_job	Crawl specific URLs for full content extraction	Slow (10–120s)	Heavy — full page content

When to use this skill

Trigger when ANY of the following is true:

• Quick fact-checking or looking up current information.
• The task asks to "search for", "look up", or "find information about" something.
• Researching trends, news, or recent developments.
• Finding local businesses, stores, or services.
• Extracting content from specific web pages for research or repurposing.
• Competitor website analysis — what are they doing?
• Price checking or availability lookup.
• The task needs current web data but doesn't require a multi-step research report.

Do NOT use this skill for:

• Deep, multi-source research reports → use design-deep-research
• Product SKU research with WooCommerce intent → use design-product-research

web_search — General Web Queries

The primary tool for quick web lookups. Configured to use Brave Search by default, with Tavily and DuckDuckGo as alternatives.

Parameters

query          — Search query (required)
max_results    — Number of results 1–10 (default: 5)
country        — ISO country code for geo-targeting (e.g., "LK" for Sri Lanka)
language       — ISO language code (e.g., "en")
freshness      — "pd" (past day), "pw" (past week), "pm" (past month), "py" (past year)
save_to_paper_store — Auto-save results (default: false)
paper_store_collection — Target collection (default: "web-search-results")
paper_store_tags — Tags for the saved record

Query Tips

Good:  "luxury fragrance market Sri Lanka 2026"
Good:  "Carolina Herrera Good Girl price"
Good:  "best Instagram engagement rate benchmarks retail 2026"
Bad:   "tell me about perfumes" (too vague)
Bad:   "what is the best perfume for men in sri lanka that is affordable
        and long lasting and comes in a nice bottle" (too long — be specific)

Geo-Targeting for Sri Lanka

When researching The Parfumerie's market:

web_search {
  query: "online perfume shopping Sri Lanka trends"
  country: "LK"
  language: "en"
}

Trend Research

Use freshness to focus on recent developments:

web_search {
  query: "fragrance trends 2026"
  freshness: "pm"  // past month
  max_results: 10
}

brave_local_search — Local Discovery

Search for businesses, places, and services with location context.

Parameters

query   — Search query (e.g., "perfume shops Colombo")
count   — Results 1–20 (default: 5)

When to use

• Finding competitor locations
• Researching local market presence
• Discovering potential stockists or partners
• Location-based content ("best perfume shops in Colombo")

Example

brave_local_search {
  query: "luxury perfume shops Colombo Sri Lanka"
  count: 10
}

run_crawl4ai_job — Deep Page Crawling

Extract full content from one or more URLs. Use when you need to read and analyze a competitor's website, extract product details from a page, or ingest a long article for repurposing.

Parameters

urls                 — Array of URLs to crawl (required, or use single `url`)
url                  — Single URL (alternative to urls)
priority             — Job priority 0–100 (default: not set)
options              — Additional Crawl4AI options (crawler config, hooks)
wait_for_completion  — Poll until done (default: false, returns immediately)
poll_interval        — Seconds between polls (default: 3, max: 30)
timeout              — Max wait time in seconds (default: 120, max: 600)

Modes

Mode	When	How
Fire-and-forget	Crawling many pages, not urgent	wait_for_completion: false → returns job ID
Wait for result	Need content immediately, 1–2 URLs	wait_for_completion: true

Use Cases

1. Competitor product page analysis:
   run_crawl4ai_job {
     url: "https://competitor.lk/product-page"
     wait_for_completion: true
   }

2. Research article ingestion:
   run_crawl4ai_job {
     urls: ["https://example.com/fragrance-trends-2026",
            "https://example.com/luxury-market-sri-lanka"]
     wait_for_completion: true
     timeout: 300
   }

3. Bulk competitor scan (background):
   run_crawl4ai_job {
     urls: ["url1", "url2", "url3", ...]
     wait_for_completion: false
   }

WP-CLI Commands

Key mcp-ai WP-CLI commands for web research workflows. Run via terminal:

Command	Use for
wp mcp-ai harness	Semantic search across research outputs and crawled content
wp mcp-ai bulk	Batch crawl and index URLs via Crawl4AI
wp mcp-ai log	View crawl job logs, errors, and timeout diagnostics

Choosing the Right Search Tool

Need a quick answer?                    → web_search
Need local business info?               → brave_local_search
Need to read a specific web page?       → run_crawl4ai_job
Need a comprehensive report?            → deep_research (design-deep-research)
Need product specs and WooCommerce copy? → research_product (design-product-research)

Paper Store Integration (on request)

When explicitly asked to persist search results:

paper_store_write {
  collection: "web-search-results"
  id: "search-topic-slug"
  title: "Search: [Query]"
  connection_id: "conn_8srcad8zylhe"
  tags: ["web-search", "topic-tag"]
  body: {
    query: "..."
    results: [...]
    searched_at: "YYYY-MM-DD"
  }
}

Critical Rules

• Use the lightest tool for the job — web_search for quick facts, deep_research for reports, never the other way around.
• Geo-target when relevant — use country: "LK" for Sri Lanka-specific searches.
• Use freshness for trends — freshness: "pw" or "pm" for current information.
• Don't crawl unnecessarily — run_crawl4ai_job is heavyweight. Use it only when you need the full content of a specific page.
• Save results when asked — Paper Store persistence only on explicit request.

Common Mistakes

WRONG:
  ● Using deep_research for "What's the weather?" (use web_search)
  ● Crawling 50 URLs with wait_for_completion: true (will time out)
  ● Not geo-targeting local searches (missing Sri Lanka results)
  ● Searching without freshness for trends (getting 2024 results in 2026)

RIGHT:
  ✅ web_search for quick facts, current events, trend checks
  ✅ run_crawl4ai_job with fire-and-forget for bulk crawls
  ✅ country: "LK" for Sri Lankan market research
  ✅ freshness: "pm" for trend monitoring

What This Skill Does NOT Cover

• Deep multi-source research reports with citations → design-deep-research
• Product SKU research with WooCommerce intent → design-product-research
• SEO keyword research and competitive SERP analysis → design-seo-content
• Converting research findings into WordPress drafts → design-content-research
• Persistent storage and structured tagging of research data → paper_store_* tools
• Brand or market trend analysis with strategic recommendations → design-deep-research

Cross-references

• Run design-deep-research for comprehensive multi-source research — use deep_research.
• Run design-product-research for product SKU lookups — use research_product.
• Run design-content-research to convert crawled content into WordPress drafts.
• Run design-seo-content to apply search insights to product page optimization.
• Run design-analytics-reporting to research industry benchmarks for performance comparison.
• Use paper_store_* with connection_id to persist search results to specific sites.