websearch-deep

Decompose complex questions into sub-questions and synthesize ranked evidence with citations.

6|1|Updated Sep 1, 2025

Search Tags:#architecture #multi-query #deep-research #citation transparency #evidence synthesis

Ready to run

npx skills add https://github.com/thomasholknielsen/claude-code-config --skill websearch-deep

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: websearch-deep
Source: https://github.com/thomasholknielsen/claude-code-config/tree/main/skills/websearch-deep
Command: npx skills add https://github.com/thomasholknielsen/claude-code-config --skill websearch-deep

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides comprehensive deep research methodology for complex, multi-faceted questions requiring synthesis across multiple domains. Implements problem decomposition, multi-query strategies, evidence synthesis, citation transparency, and iterative refinement.

Core Features & Use Cases

Problem Decomposition: Break complex questions into 3-5 sub-questions.
Multi-Query Variations: 3-5 queries per sub-question for coverage.
Evidence Synthesis: Rank and summarize sources with citations.
Iterative Refinement: Up to 5 iterations to reach completeness.

Quick Start

Use for architecture decisions, technology selection, and strategic analyses; expect a multi-source, evidence-based output.

Frequently Asked Questions about websearch-deep

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

FAQPage Schema

How do I conduct deep research across multiple domains to answer complex questions?▼

Deep research uses problem decomposition to break complex questions into 3-5 sub-questions, then generates 3-5 query variations per sub-question to gather comprehensive evidence. Sources are ranked, synthesized with numbered citations, and iteratively refined up to 5 times to reach completeness and evidence transparency.

What's the best way to compare technologies and make architecture decisions based on evidence?▼

Structured research decomposes the decision into sub-questions, searches multiple angles per question, ranks and synthesizes source evidence with citations, then iteratively refines findings. This methodology produces a verification artifact documenting the research execution and evidence basis for your choice.

Can I use multi-query research to evaluate strategic options across different domains?▼

Yes. Multi-query deep research applies to multi-domain investigations by decomposing strategic questions, generating varied queries for each sub-question, synthesizing ranked evidence, and refining iteratively. The approach supports technology selection, architecture decisions, and comparative analyses requiring cross-domain synthesis.

How do I ensure my research findings are transparent and well-sourced?▼

Evidence synthesis with numbered citations documents source rankings and reasoning at each step. Iterative refinement up to 5 times strengthens weak areas, and a verification artifact records the full research execution, source evaluation, and synthesis logic for reproducibility and transparency.

What preparation do I need before starting a deep research project?▼

Clarify your complex question and identify its domain scope. Deep research works best when the core question is decomposable into 3-5 sub-questions; no special tools or environment are required. The methodology handles problem breakdown, query generation, evidence synthesis, and iterative refinement from there.

When should I avoid using multi-query deep research for investigation?▼

Deep research suits complex, multi-domain questions requiring evidence synthesis and citation transparency. It may be overkill for simple factual lookups or single-domain queries with obvious answers. Use it when you need comprehensive cross-domain synthesis, strategic comparison, or architecture justification.