跳到正文
TakoAPI
返回技能列表

academic-deep-research

由 kesslerio 提供

Search & Research
10.8k下载量由 clawskills.sh 提供
38星标
20浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Transparent, rigorous research with full.

描述

该技能的功能

Conducts multi-phase academic research using a fixed methodology: two full research cycles per theme, APA 7th edition citations, evidence hierarchy annotations, and three mandatory user checkpoints before the final report. Runs entirely on native OpenClaw tools without external API dependencies.

The mandated two-cycle structure and explicit evidence hierarchy force gap identification and source cross-referencing that single-pass AI research tools skip entirely.

何时使用

  • Literature review for a graduate thesis on a contested scientific topic
  • Competitive intelligence report on market players before a product launch
  • Fact-checking a news article's claims against primary sources
  • Synthesizing research on a medical treatment before a patient consultation
  • Trend analysis across multiple domains for a strategic planning document

示例工作流

输入

User asks: What does current research say about muscle mass loss during GLP-1 weight loss treatment?

智能体
  1. 1

    Ask 2-3 clarifying questions about depth, scope, and source preferences, then wait for user response

  2. 2

    Present a structured research plan with 3-5 themes, tool sequence, and expected deliverables, then wait for approval

  3. 3

    Run two web_search cycles (count=20) and web_fetch on primary sources for each theme, documenting hypothesis evolution and contradictions between cycles

  4. 4

    Synthesize cross-theme patterns, resolve or flag contradictions, and annotate conclusions with confidence levels

  5. 5

    Deliver a full narrative report in APA 7th edition with in-text citations and a complete reference list

输出

A multi-section academic report with flowing prose, integrated citations, a contradictions analysis, a limitations section, and an alphabetized APA reference list

安装

clawhub install kesslerio/academic-deep-research

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接