Is the final output provided in the chat window?

No, it generates a structured, standalone Markdown report file in a dedicated directory to ensure the final product is professional and easy to navigate.

Can I control the model settings for child processes?

By default, it maintains your existing model settings to ensure consistency and prevent unexpected costs, only adjusting parameters if you explicitly authorize it.

What tools does it use for web searching?

The skill prioritizes installed skills first, followed by MCP tools like Firecrawl and Tavily, falling back to direct fetching via curl or wget only when necessary.

Is it secure to run these agents?

Yes, child processes are executed within a restricted 'workspace-write' sandbox by default, minimizing permissions while providing the necessary environment for research.

How does this skill handle large research tasks?

It splits the main objective into parallel sub-goals, executing them simultaneously via child processes to significantly speed up data collection and analysis.

Deep Research Orchestrator

Name: Deep Research Orchestrator
Author: Fancu1

byFancu1

0•

Web Scraping y Recopilación de Datos

Orchestrates multi-agent parallel workflows to perform comprehensive web research, data collection, and structured report generation.

The Deep Research skill transforms complex investigative objectives into an automated, parallelized production pipeline by decomposing high-level goals into discrete sub-tasks. It manages independent sub-processes within secure sandboxes, intelligently leveraging web scraping tools and MCP integrations to gather evidence across multiple domains simultaneously. By programmatically aggregating findings into a unified, polished Markdown report rather than chat snippets, it provides a professional-grade research environment ideal for competitive analysis, technical deep-dives, and systematic documentation retrieval.

Características Principales

01Priority-based tool selection leveraging Skills and MCP (Firecrawl, Tavily)

02Intelligent goal decomposition into manageable, parallel sub-targets

030 GitHub stars

04Programmatic result aggregation with chapter-by-chapter iterative refinement

05Multi-agent parallel task orchestration using nested codex execution

06Sandbox-isolated execution with automated logging and file-based delivery

Casos de Uso

01Large-scale web scraping and distributed data collection across diverse sources

02Comprehensive competitive intelligence and industry market analysis

03Systematic technical documentation, literature reviews, and evidence synthesis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fancu1/dotfile deep-research

For use in Claude.ai and ChatGPT

Download Skill