deep-research

Coordinate multi-source web research via an isolated background agent.

5|Updated Feb 24, 2026
One-click install
npx skills add https://github.com/mrtolkien/GHOST --skill deep-research-mrtolkien
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/mrtolkien/GHOST/tree/main/prompts/skills/deep-research
Command: npx skills add https://github.com/mrtolkien/GHOST --skill deep-research-mrtolkien

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill handles multi-source web research by delegating heavy fetching to an isolated background agent, protecting your ongoing conversation from context bloat.

Core Features & Use Cases

  • Spawns an isolated research agent to perform web fetches and produce a concise summary.
  • Uses knowledge_search first to reuse existing notes, and employs reference-import for importing external documentation when available.
  • Generates structured findings reports via the report_findings tool for later review.

Quick Start

Ask the operator a multi-source research question and trigger the deep-research agent to begin background reading.

Frequently Asked Questions about deep-research

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

FAQPage Schema
How do I perform multi-source web research without bloating my conversation context?

Multi-source web research can be performed without context bloat by spawning an isolated background agent to handle heavy fetching. This agent reads multiple web pages and returns a concise summary to your main conversation.

What is the best way to coordinate cross-source comparisons for complex evaluations?

The best way to coordinate cross-source comparisons is using an agent-control workflow to manage a deep-research agent. This agent reads several pages, applies knowledge checks, and produces a structured report for multi-factor decisions.

How do I reuse existing notes before fetching new information from the web?

You can reuse existing notes by performing a knowledge_search before web fetching. The research workflow checks your knowledge base first and employs reference-import for external documentation when available.

Does this research agent handle simple web lookups inline?

No, the research agent avoids inline fetching for simple lookups. It specifically applies to questions requiring cross-source comparisons, recommendations, evaluations, or multi-factor decisions where reading several pages is needed.

How do I generate a structured findings report after background web research?

You generate a structured findings report by using the report_findings tool after the background agent completes its web fetching. This produces a concise summary of the research for later review.