deepsearch

Orchestrate multi-round web searches into a structured Markdown research report.

417|87|Updated Nov 12, 2025
One-click install
npx skills add https://github.com/bytedance/agentkit-samples --skill deepsearch-bytedance
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: deepsearch
Source: https://github.com/bytedance/agentkit-samples/tree/main/skills/deepsearch
Command: npx skills add https://github.com/bytedance/agentkit-samples --skill deepsearch-bytedance

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

DeepSearch provides a workflow-driven approach to perform deep, multi-turn web research and synthesis using LLMs, reducing manual search coordination and ensuring structured, evidence-backed outputs.

Core Features & Use Cases

  • Workflow-driven deep research: orchestrates multi-round web searches and iterative analysis.
  • LLM-powered synthesis: aggregates findings into a comprehensive report.
  • Traceable decisions: avoids repeated topics and tracks progress for reproducibility.
  • Use Case: When you need to generate a detailed, up-to-date analysis of a complex topic using real-time web information.

Quick Start

Initiate a multi-round deep search on a topic with a specified depth and generate a final Markdown research report.

Frequently Asked Questions about deepsearch

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

FAQPage Schema
How do I automate multi-round web research to generate a structured report?▼

Multi-round web research is automated by orchestrating iterative web searches and LLM-powered synthesis to produce a structured Markdown report. It tracks topics across rounds to avoid repetition and aggregates findings into a comprehensive final output.

What is the best way to synthesize complex topics requiring up-to-date information?▼

Synthesizing complex topics with up-to-date information is best handled through a workflow-driven deep search that uses iterative analysis. This process aggregates real-time web search results into a comprehensive, evidence-backed Markdown report.

Can I track research progress and decisions across multiple search rounds?▼

Tracking research progress across multiple search rounds is supported via a JSON-driven decision workflow. It monitors topics and findings iteratively, ensuring traceable decisions and preventing repeated topics for reproducible research outputs.

How do I configure search depth for iterative web research?▼

Search depth for iterative web research is configurable when initiating the query. You can specify the desired depth to control the intensity of the multi-round web searches before generating the final Markdown research report.

When do I need a workflow-driven deep research tool instead of a single web search?▼

A workflow-driven deep research tool is needed for complex topics requiring comprehensive synthesis and up-to-date information. Unlike a single web search, it coordinates multi-round searches and iterative topic refinement to yield clear, evidence-backed reports.

What format does the multi-round deep search output use for the final report?▼

The multi-round deep search outputs the final report in Markdown format. This structured document aggregates the synthesized findings from the iterative web searches, providing a clear and comprehensive research summary.