How does deep-research differ from standard web search skills?

Standard tools like brave-research are for quick, single-topic lookups. Deep-research is designed for complex investigations requiring 3+ independent sub-topics, parallel processing, and consolidated multi-source reporting.

Does it support research in multiple languages?

Absolutely. It is optimized to search across global sources (including English and Japanese) and synthesize them into a single, cohesive report in your preferred language.

Can I control the cost and speed of the research?

Yes. The skill includes a 'compare-lite' mode for fast, low-cost comparisons and allows you to specify model preferences (e.g., 'cost-focused' or 'premium quality') to adjust the research depth.

Where does the skill store the research results?

The skill automatically creates a structured directory at /tmp/deep-research/{slug}/ where it saves individual agent outputs and the final consolidated report.md file.

Deep Research

Name: Deep Research
Author: haboshi

byhaboshi

0•

网络抓取与数据收集

Performs comprehensive multi-agent web research and generates structured analysis reports across technical, market, and academic domains.

Deep Research is an advanced multi-agent capability for Claude that automates complex information gathering and synthesis. Unlike standard search tools that provide quick lookups, this skill breaks down broad queries into multiple sub-topics, deploys parallel search agents to investigate independent sources, and synthesizes findings into high-quality, structured reports. It features two distinct operating modes—compare-lite for rapid competitive analysis and deep-full for exhaustive market or technical investigations—enabling Claude to handle everything from 'A vs B' comparisons to full-scale industry research with automated citations and quality metadata.

主要功能

01Smart model policy that balances cost-effective searching with high-reasoning synthesis using GPT or Claude models

02Parallel multi-agent searching for high-speed information gathering across multiple sub-topics

030 GitHub stars

04Specialized agent types for Technical, Market, User Experience, Academic, Policy, and Creative domains

05Optimized modes: Compare-Lite for rapid benchmarking and Deep-Full for exhaustive 3-round investigation

06Automated structured report generation in Markdown with source verification and quality scoring

使用场景

01Technical benchmarking and comparative analysis of frameworks, tools, or services

02Gathering multi-perspective insights for strategic planning and new product ideation

03In-depth market research, trend analysis, and competitor landscape mapping

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add haboshi/claude-code-skills deep-research

For use in Claude.ai and ChatGPT

主要功能

01Smart model policy that balances cost-effective searching with high-reasoning synthesis using GPT or Claude models

02Parallel multi-agent searching for high-speed information gathering across multiple sub-topics

030 GitHub stars

04Specialized agent types for Technical, Market, User Experience, Academic, Policy, and Creative domains

05Optimized modes: Compare-Lite for rapid benchmarking and Deep-Full for exhaustive 3-round investigation

06Automated structured report generation in Markdown with source verification and quality scoring

使用场景

01Technical benchmarking and comparative analysis of frameworks, tools, or services

02Gathering multi-perspective insights for strategic planning and new product ideation

03In-depth market research, trend analysis, and competitor landscape mapping