deep-research

Orchestrate multi-agent research into parallel subgoals and a structured report.

2|Updated Mar 25, 2026
One-click install
npx skills add https://github.com/billlzzz26/note-taker-convex-ai-convex --skill deep-research-billlzzz26
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/billlzzz26/note-taker-convex-ai-convex/tree/main/.agents/skills/deep-research
Command: npx skills add https://github.com/billlzzz26/note-taker-convex-ai-convex --skill deep-research-billlzzz26

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a self-contained workflow to orchestrate deep research using multiple agents, breaking a research target into parallel subgoals, running subprocesses in non-interactive mode, and delivering a polished final report rather than chat fragments. It ensures results are collected, structured, and delivered as independent artifacts with traceable logs.

Core Features & Use Cases

  • Parallel subgoal decomposition and orchestration by a central controller.
  • Sub-processes gather, extract, and structure information into Markdown artifacts.
  • Final deliverable is an independent product file containing conclusions and evidence, with links to sources.
  • Automated aggregation via scripts and iterative polishing by chapters for a publish-ready report.
  • Use cases include systematic web research, competitive analysis, batch link/data chunk retrieval, and long-form writing with evidence integration.

Quick Start

Split the target into parallel subgoals, run each with claude -p in non-interactive mode, and aggregate results into a final report.

Frequently Asked Questions about deep-research

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

FAQPage Schema
How do I automate deep web research and generate a structured report?

Automated deep web research uses a multi-agent workflow to decompose research targets into parallel subgoals, execute sub-queries, and aggregate results into a structured final report artifact.

What is multi-agent workflow orchestration for competitive analysis?

Multi-agent workflow orchestration for competitive analysis uses a central controller to split research into parallel sub-processes, gather data, and aggregate findings into an independent product file with traceable logs.

How to run Claude Code in non-interactive mode for parallel research subgoals?

Run Claude Code in non-interactive mode by executing sub-queries with the `claude -p` command, allowing a central controller to orchestrate parallel subgoals and aggregate results via scripts.

Does multi-agent web research require installed skills or MCP servers?

Multi-agent web research prioritizes installed skills before MCP servers, operating strictly in non-interactive mode to ensure sub-processes gather and structure information independently.

What is the best way to collect and aggregate data into a Markdown artifact?

The best way to collect and aggregate data into a Markdown artifact is using a multi-agent workflow that runs sub-processes for data extraction and uses scripts for automated aggregation and iterative polishing.

Why use non-interactive subprocesses for long-form writing and report generation?

Non-interactive subprocesses for long-form writing ensure that parallel subgoals execute without interruption, gathering evidence and aggregating results into a publish-ready report with traceable logs rather than chat fragments.