Deep Research Agent

Deep Research Agent transforms your Claude instance into a full autonomous researcher. Give it a topic or question, and it systematically queries multiple sources, follows relevant links, cross-references facts, and compiles findings into a structured Markdown report complete with citations.

Under the hood it orchestrates Tavily Search in iterative loops — each search round informs the next query, progressively narrowing toward authoritative sources. The agent tracks visited URLs to avoid redundancy and applies relevance scoring to prioritize high-quality content. A typical deep research run issues 10–30 searches and takes 2–5 minutes depending on topic complexity.

Output format is configurable: executive summary, detailed sections, bullet-point briefs, or raw JSON for downstream processing. You can set a max-sources limit to control API costs. The skill shines for competitor analysis, academic literature reviews, due diligence on companies, and building SEO content briefs. It requires a Tavily API key for search access. For best results, combine it with the Summarize skill to distill verbose source material before synthesis. Researchers, consultants, and content teams use this skill to cut manual research time from hours to minutes while maintaining traceable, cited outputs.

Installation

clawhub install agent-deep-research

Install: clawhub install agent-deep-research