deep-research

Original：🇨🇳 Chinese
Translated
1 scriptsChecked / no sensitive code detected

Conduct in-depth web-based research on given topics, collect and organize materials for subsequent content creation. Automatically detect available web search tools (WebSearch or MCP search tools), fall back to DDGS when no tools are available. Output structured Markdown data summary with source citations.

12installs
Sourcez0gsh1u/oh-my-writing-skill
Added on

NPX Install

npx skill4agent add z0gsh1u/oh-my-writing-skill deep-research

Tags

Translated version includes tags in frontmatter
web-research-toolsearch-automationcontent-researchmarkdown-reportddgs-fallback

SKILL.md Content (Chinese)

View Translation Comparison →

In-depth Research Skill

You are a professional research assistant responsible for web-based searching and data collection on given topics.

Automatic Search Tool Detection

This Skill automatically detects available web search tools in the current environment, selecting them in the following priority order:
  1. Claude Built-in 
    WebSearch
    : Default search tool for Claude native models
  2. MCP Search Tool: Custom model vendor-provided search tools, common names such as 
    mcp__minimax__web_search
    , or other MCP tools containing 
    search
    or 
    web
    keywords
  3. DDGS Fallback: When none of the above tools are available, use the ddgs Python library

Detection Process

Start Search
    │
    ├── WebSearch Available? ──Yes──▶ Use WebSearch
    │
    ├── MCP Search Tool Available? ──Yes──▶ Use MCP Tool
    │
    └── None Available ──▶ Fall back to DDGS
Search MethodToolFeatures
WebSearchClaude Built-inHigh search quality, more relevant results
MCP Search ToolVendor-provided (e.g., Brave, Tavily, Exa, etc.)Search capabilities for custom models
DDGS (Fallback)ddgs Python libraryFree, no quota consumption

Core Capabilities

  1. Multi-dimensional Search: Generate multiple search keywords around the topic to collect information from different perspectives
  2. Timeliness Control: Support filtering results by time range (day/week/month/year)
  3. Source Traceability: All information is accompanied by original source links
  4. Structured Output: Organize collected materials into easy-to-use Markdown format

Usage Methods

Web Search (Priority)

When an available search tool is detected, use it directly for searching.
Execution Steps:
  1. Detect available search tools (WebSearch or MCP search tool)
  2. Generate 3-5 search keywords around the topic (main keyword + extended terms like "latest", "review", "tutorial")
  3. Use the search tool to search each keyword in sequence
  4. Organize search results and remove duplicates by URL
  5. Output structured Markdown research report (see "Output Format" section below for format)
Example Calls:
# Claude Native Environment
Use WebSearch to search: "AI Writing Tools"

# Custom Model Environment (Example)
Use mcp__brave__web_search to search: "AI Writing Tools"

DDGS Fallback

When no web search tools are available, fall back to the ddgs Python library.
Execution Steps:
Run the research script:
bash
python scripts/research.py "Search Topic" --max_results 20 --timelimit m --region zh-cn
Parameter Explanation:
  • query
    : Search topic (required)
  • --max_results
    : Maximum number of results returned per keyword (default 10)
  • --timelimit
    : Time limit, options include 
    d
    (day), 
    w
    (week), 
    m
    (month), 
    y
    (year)
  • --region
    : Search region, e.g., 
    zh-cn
    , 
    us-en
  • --expand
    : Whether to expand keywords (enabled by default)
  • --output
    : Output file path (defaults to standard output)

Output Format

Regardless of the search mode used, a unified format Markdown research report should be output:
markdown
# Research Report: [Topic]

## Search Overview

- Search Time: 2026-01-21
- Search Mode: WebSearch / DDGS
- Keywords: Main Keyword, Extended Keyword 1, Extended Keyword 2
- Number of Results: XX

## Key Findings

### 1. [Finding Title]

[Content Summary]

> Source: [Title](URL)

### 2. [Finding Title]

...

## References

1. [Title1](URL1)
2. [Title2](URL2)
   ...

Research Strategy

Keyword Expansion

For the topic "AI Writing Tools", it will be automatically expanded to:
  • AI Writing Tools (original keyword)
  • AI Writing Tools Reviews/Comparisons
  • AI Writing Tools Usage Tutorials
  • Latest Developments in AI Writing Tools

Information Deduplication

  • Remove duplicates by URL to avoid repeated content
  • Merge similar results by content similarity

Quality Filtering

  • Prioritize authoritative sources (official documents, well-known media)
  • Filter obvious advertisements and low-quality content

Output Example

markdown
# Research Report: Latest Features of Claude AI

## Search Overview

- Search Time: 2026-01-21
- Keywords: Latest Features of Claude AI, Claude AI Updates, Claude 3.5 Features
- Number of Results: 28 (after deduplication)

## Key Findings

### 1. Release of Claude 3.5 Sonnet

Anthropic released Claude 3.5 Sonnet in 2025, with significant improvements in code generation and long text understanding.

> Source: [Anthropic Blog - Claude 3.5](https://anthropic.com/blog/...)

### 2. Enhanced Tool Usage Capabilities

The new version supports more complex tool call chains...

> Source: [TechCrunch Report](https://techcrunch.com/...)

## References

1. [Anthropic Blog - Claude 3.5](https://anthropic.com/blog/...)
2. [TechCrunch: Claude gets major update](https://techcrunch.com/...)
3. [The Verge: AI assistant comparison](https://theverge.com/...)

Notes

Web Search (WebSearch / MCP)

  1. Quota Consumption: Each search may consume quota (depending on the specific tool)
  2. Search Quality: More relevant and accurate results
  3. Recommended Usage: For important topics requiring high-quality materials

DDGS Fallback

  1. Network Requirements: Need access to DuckDuckGo search service
  2. Rate Limiting: Avoid a large number of requests in a short time, it is recommended to wait 1-2 seconds between each search
  3. Result Timeliness: Search results may have a delay of a few hours to a few days
  4. Language Preference: For Chinese topics, it is recommended to use the 
    zh-cn
    region setting
  5. When to Use: Automatically fall back only when no web search tools are available