利用claude code agent框架一步一步实现deep research！很强大很简单的skills。我一步一步介绍实现deep research，因为deep research就是agent框架第一应用，对比一下各个框架实现这个deep research，就知道哪个框架才是真厉害。
liangdabiao/Claude-Code-Deep-Research-main
Skill Claude Code
A research-report checking skill that reviews whether claims have accurate, trustworthy, and properly formatted citations. A citation is a source reference showing where a fact came from.
liangdabiao/Claude-Code-Deep-Research-main
Skill Claude Code
A controller for a research graph, where connected steps represent different parts of an investigation. It can generate, combine, refine, and score research paths.
liangdabiao/Claude-Code-Deep-Research-main
Skill Claude Code
A question-planning skill that turns a rough research question into a clearly scoped research task and structured prompt. It asks follow-up questions when details are missing.
liangdabiao/Claude-Code-Deep-Research-main
Skill Claude Code
A skill that runs a seven-stage deep-research process. It takes a structured research task, sends parts of it to several agents in parallel, and produces a cited report.
liangdabiao/Claude-Code-Deep-Research-main
Skill Claude Code
A research-writing skill that combines findings from multiple research agents into one structured report. It handles disagreements and extracts shared conclusions.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: