deep-research

Synthesize evidence from 10+ sources into a citation-backed Markdown research report.

Updated Mar 16, 2026
One-click install
npx skills add https://github.com/Zhushuaijiang/auto-company-win --skill deep-research-zhushuaijiang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/Zhushuaijiang/auto-company-win/tree/main/.agents/skills/deep-research
Command: npx skills add https://github.com/Zhushuaijiang/auto-company-win --skill deep-research-zhushuaijiang

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Conduct enterprise-grade research with multi-source synthesis, rigorous verification, and citation-backed reporting to replace manual, time-consuming analysis tasks.

Core Features & Use Cases

  • 8-phase research pipeline (Scope → Plan → Retrieve → Triangulate → Synthesize → Critique → Refine → Package) with automated validation and credibility scoring
  • Multi-depth modes: quick, standard, deep, and ultradeep to balance speed and depth
  • Parallel information gathering: automated parallel web searches and agent-driven deep dives
  • Progressive context management: on-demand loading of references and templates to optimize latency
  • Output fidelity: generates Markdown reports with a complete bibliography, methodology appendix, and a reproducible file structure
  • Autonomy by design: operates with minimal user interaction, providing plan, execution, and delivery without blocking prompts

Real-world use: A product leader wants a 10+ source, evidence-backed analysis comparing multiple strategies and architectures. The skill autonomously fetches sources, triangulates claims, validates citations, and delivers a comprehensive report ready for executive review.

Quick Start

Initiate a deep research run on your topic to generate a citation-backed report.

Frequently Asked Questions about deep-research

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a citation-backed research report from multiple web sources autonomously?

Autonomous multi-source research uses an 8-phase pipeline to retrieve, triangulate, and synthesize evidence from 10+ diverse sources, delivering a structured Markdown report with a bibliography and methodology appendix saved under ~/.claude/research_output/.

How does parallel retrieval work for deep research and evidence triangulation?

Parallel retrieval executes automated web searches and agent-driven deep dives concurrently to gather evidence. The pipeline then triangulates claims across 10+ diverse sources to validate citations and score credibility before synthesizing the final report.

Can I balance research speed and depth for different analysis tasks?

You can balance speed and depth using four multi-depth modes: quick, standard, deep, and ultradeep. These modes scale parallel information gathering and automated validation to match your analysis latency requirements.

What is the best way to validate citations and score source credibility in automated research?

The best way to validate citations is using the research pipeline's automated validation and credibility scoring. It cross-references claims during the triangulation phase and appends a methodology appendix to ensure evidence-backed reproducibility.

Do I need to manually intervene during the multi-source synthesis process?

No manual intervention is needed during multi-source synthesis. The skill operates autonomously by design, executing the plan, parallel retrieval, and packaging without blocking user prompts or requiring continuous manual interaction.

What is included in the structured Markdown output for executive review?

The structured Markdown output includes a comprehensive research report, a complete bibliography of the 10+ diverse sources consulted, and a methodology appendix detailing the validation process, ready for executive review.