deep-research

Generate evidence-backed research reports with cited claims and confidence scoring.

Updated Feb 10, 2026
One-click install
npx skills add https://github.com/monktui/git-review --skill deep-research-monktui
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/monktui/git-review/tree/main/.claude/skills/deep-research
Command: npx skills add https://github.com/monktui/git-review --skill deep-research-monktui

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill turns vague or complex questions into a structured, evidence-backed research report by coordinating multiple parallel agents and synthesizing their findings.

Core Features & Use Cases

  • Guided clarification interview: It asks targeted questions to refine scope, constraints, and source preferences before any research starts.
  • Parallel, depth-based investigation: It decomposes the query into non-overlapping subtopics (3/5/10 researchers based on --depth) and runs web research in parallel for each subtopic.
  • Credibility scoring and confidence synthesis: It collects JSON-formatted findings, deduplicates facts and sources, cross-validates claims, and assigns confidence indicators with a final confidence statistics summary.
  • Practical output format: It produces a comprehensive report with executive summary, key findings, themed analysis, and categorized sources.

Quick Start

Use the deep-research skill to research the question "What are the latest best practices for securing a Rust web service against common web vulnerabilities?" with default depth.

Frequently Asked Questions about deep-research

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

FAQPage Schema
How do I generate an evidence-backed research report for an open-ended technical question?

To generate an evidence-backed research report, you can use a guided clarification process to refine scope, decompose the query into subtopics, and run parallel web research to synthesize findings with confidence scoring.

What is parallel research planning and how does it work for complex topics?

Parallel research planning decomposes a complex query into non-overlapping subtopics, assigns multiple agents to investigate each one simultaneously via web search, and cross-validates the extracted JSON evidence to deduplicate facts and sources.

How do I structure web search findings with source credibility scoring?

You structure web search findings by extracting claims into a JSON format, tiering source credibility across parallel investigations, and cross-validating claims to produce a final report with explicit confidence indicators.

Can I use depth-controlled decomposition for broad market research tasks?

Yes, depth-controlled decomposition supports broad market research by scaling the investigation across 3, 5, or 10 subtopics, ensuring comprehensive coverage of recent developments and explicit limitations across multiple areas.

What is the best way to synthesize cited claims from multiple subtopics?

The best way to synthesize cited claims is to orchestrate parallel agents for each subtopic, collect structured JSON findings, deduplicate overlapping facts, and cross-validate evidence to generate a final themed analysis.

When should I not use parallel agents for web research?

You should avoid using parallel agents for web research if your task lacks the breadth requiring multiple subtopic decomposition, or if immediate, single-source answers are sufficient without explicit confidence and credibility synthesis.