deep_research

Agent Skills › siddsachar/row-bot › deep_research

deep_research

GitHub

指导Agent进行深度主题研究，通过检查记忆、多源搜索及学术文献检索获取信息，最终生成包含摘要、关键发现及来源引用的结构化研究报告，并保存结果至记忆。

bundled_skills/deep_research/SKILL.md siddsachar/row-bot

触发场景

深入调研某个主题撰写研究报告彻底调查某事

安装

npx skills add siddsachar/row-bot --skill deep_research -g -y

更多选项

非标准路径

npx skills add https://github.com/siddsachar/row-bot/tree/main/bundled_skills/deep_research -g -y

不安装直接使用

npx skills use siddsachar/row-bot@deep_research

指定 Agent (Claude Code)

npx skills add siddsachar/row-bot --skill deep_research -a claude-code -g -y

安装 repo 全部 skill

npx skills add siddsachar/row-bot --all -g -y

预览 repo 内 skill

npx skills add siddsachar/row-bot --list

SKILL.md

Frontmatter

{
    "icon": "🔬",
    "name": "deep_research",
    "tags": [
        "research",
        "analysis"
    ],
    "author": "Row-Bot",
    "version": "1.1",
    "activation": {
        "phrases": [
            "deep research",
            "research competitors",
            "competitor research",
            "structured report",
            "multi-source research",
            "look around",
            "real story",
            "investigate thoroughly"
        ],
        "examples": [
            "I need to research competitors and produce a structured report"
        ],
        "keywords": [
            "research",
            "competitors",
            "sources",
            "report",
            "market",
            "compare",
            "investigate"
        ],
        "negative_phrases": [
            "meeting notes",
            "action items",
            "human tone"
        ]
    },
    "description": "Perform multi-source research on a topic and produce a structured report.",
    "display_name": "Deep Research",
    "enabled_by_default": true
}

When the user asks you to research a topic in depth, write a research report, or investigate something thoroughly, follow these steps:

Check Existing Knowledge — Before searching the web, review your recalled memories and use search_memory for the topic. The user may have researched this before, or you may already have saved facts, sources, or prior findings. Build on what you already know rather than starting from scratch.
Clarify Scope — If the topic is broad, ask one focused question to narrow it down before proceeding. Otherwise, proceed directly.
Initial Search — Run 2–3 web searches with varied queries to gather diverse perspectives on the topic.
Source Deep-Dive — Pick the 3–5 most promising URLs from the search results and read their full content using the URL reader.
Academic Check — If the topic is scientific or technical, search arXiv for relevant recent papers. Summarise key findings from the top 1–2 results.
Synthesise — Compile findings into a structured report:
- Executive Summary — 2–3 sentence overview
- Key Findings — Numbered list of the most important points
- Details — Deeper discussion organised by sub-topic
- Open Questions — What remains unclear or debated
- Sources — List all URLs and papers referenced
Cite Everything — Every claim should reference its source with a numbered citation.
Save Key Findings — Save the most important facts and conclusions to memory so future research on this topic builds on today's work. Use specific entity categories (person, concept, project, fact) and link related findings together.