Quick research

Single Search

Purpose

Get fast, focused answers by crawling a small set of authoritative URLs and extracting the relevant information. No multi-agent pipeline, no planning phase, no validation stage. This is the default research mode — use it whenever you lack context about any technology.

When to Use (Auto-Detect)

Always call websearch or load this skill when:

• You are about to write code using an API/library you're not 100% confident about
• You encounter an unfamiliar function signature, parameter, or return type
• You need to verify a library version, feature availability, or deprecation status
• You need a quick code example for a specific API method
• You encounter an error and need to look up the error message or fix
• You need to confirm a configuration option, CLI flag, or environment variable
• The question can be answered from 1-3 sources
• You are short on context for any technology usage — stop and search before coding

When NOT to Use

• The topic requires comparing multiple sources or perspectives — use deep-research
• You need to understand a new framework from scratch (3+ subtopics) — use deep-research
• The question is about code in the current project — answer directly from the codebase
• The user explicitly asked for "deep research" or "comprehensive report"
• The project will need this knowledge repeatedly — research and convert to skill

Inputs

Input	Type	Required	Description
query	string	Yes	The specific question to answer
seed_urls	array of string	No	Suggested URLs to check (auto-discovered if omitted)

Expected Outputs

A concise answer with:

• Direct answer to the question
• 1-3 code examples if applicable
• Source URLs for each claim

Exact Workflow Steps

1. Determine 1-3 authoritative URLs (official docs, API refs)
2. Fetch them using webfetch
3. For JS-rendered pages or incomplete responses, use Playwright MCP tools:
   • browser_navigate to load the page
   • browser_snapshot to extract structured content via element refs
4. Extract the specific information requested
5. Return a concise answer with sources

Required Checks

• Every claim has a source URL
• Answer directly addresses the question

Failure Modes

Failure Condition	Response
URL returns 404	Try Playwright MCP as fallback
Information not found at any URL	Report "not found in searched sources"

Examples

Example: "What's the cacheLife default profile in Next.js 16?"

1. Crawl nextjs.org/docs/app/api-reference/functions/cacheLife
2. Extract: default is stale=5min, revalidate=15min, expire=never
3. Return answer with source URL

Example: "Fix: ModuleNotFoundError: No module named 'aiohttp'"

1. This is a context gap — stop and search
2. websearch("python install aiohttp package")
3. Extract: pip install aiohttp
4. Proceed with the fix

References

Reference	Path
Web Scraper skill	../web-scraper/SKILL.md