deep-research

Orchestrates multi-agent web research to produce cited, whitepaper-style reports.

Updated Nov 10, 2013
One-click install
npx skills add https://github.com/bnferguson/dotfiles --skill deep-research-bnferguson
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/bnferguson/dotfiles/tree/main/claude/config/skills/deep-research
Command: npx skills add https://github.com/bnferguson/dotfiles --skill deep-research-bnferguson

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Answering complex research questions requires gathering, cross-referencing, and synthesizing dozens of sources, which is slow and error-prone when done manually or in a single pass. This Skill decomposes a broad research question into parallel research threads and produces a structured, fully cited report. ## Core Features & Use Cases - Multi-Agent Architecture: A lead agent interviews the user and plans 10+ research threads, parallel researcher agents gather web sources into structured notes, and a report-writer agent synthesizes everything into a final report. - Structured Outputs: Generates a research plan, per-subtopic research notes, a main report with numbered citations, and a separate sources bibliography. - Research Methodology Guidance: Includes a references file with frameworks (5 W's, PESTLE, SWOT), source credibility criteria, synthesis strategies, and a report quality checklist. - Use Case: Ask for a comprehensive analysis of a technology landscape, and receive a whitepaper-style report with an executive summary, critical analysis, and a complete bibliography of cited sources. ## Quick Start Ask the agent to conduct deep research on a topic, for example: research the current state of quantum error correction and produce a comprehensive cited report.

Frequently Asked Questions about deep-research

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

FAQPage Schema
How do I run a deep research task with multiple agents?▼

Invoke the skill with a research question, and the lead agent interviews you to scope the work, then spawns 10+ parallel researcher agents via the Task tool. Each researcher saves structured notes, and a report-writer agent synthesizes them into a final cited report.

What output files does the deep research workflow generate?▼

It produces a research_plan.md file, a research_notes/ directory with one markdown file per subtopic, a [topic]-report.md final report with numbered citations, and a [topic]-sources.md bibliography describing each source.

When should I not use multi-agent deep research?▼

Avoid it for simple fact-finding, single-source lookups, quick exploratory searches, or code-only research within repositories. The multi-phase pipeline is designed for comprehensive, whitepaper-level investigations requiring many sources.

How does the research process ensure source quality and accuracy?▼

The references file defines source credibility criteria covering authority, accuracy, objectivity, currency, and coverage, plus an evidence hierarchy prioritizing peer-reviewed work. Findings are cross-referenced across researcher notes, and conflicts or gaps are explicitly documented.

Can I control the cost and speed of the research agents?▼

Yes. The skill recommends using the haiku model for researcher agents to reduce costs and the sonnet model for the report-writer to improve synthesis quality. All researcher agents launch in parallel in a single message for speed.