Back to Details

deep-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

/deep-research — Multi-source research with archive

/deep-research — 多来源研究及归档功能

Plans, executes, and synthesizes research from multiple sources. Archives the output so the corpus compounds.
规划、执行并整合来自多渠道的研究内容。归档输出结果，使研究 corpus 不断积累。

Step 1 — Frame the question

步骤1 — 明确研究问题

Restate the research question in one tight sentence. If ambiguous, ask the user:
  • What's the decision this research will inform?
  • What's the minimum useful answer? (Saves over-researching.)
  • Any sources to prioritize or avoid?
Output: 
**Research question:** <one sentence>
用简洁的一句话重述研究问题。若问题模糊，向用户询问：
  • 本次研究将为哪项决策提供依据？
  • 最基础的有效答案是什么？（避免过度调研）
  • 是否有需要优先参考或规避的信息来源？
输出格式：
**研究问题:** <一句话>

Step 2 — Plan the sources

步骤2 — 规划信息来源

Pick from this menu based on the question type. Note which sources you'll hit and why.
SourceWhen to useTool
Web search (Google)Authoritative articles, docs, official statements
WebSearch
/last30days
What people are actually saying right now — Reddit, X, YouTube, HN, web recency
Skill({skill: "last30days", args: "<topic>"})
Specific URLsWhen the user hands over starting URLs
WebFetch
Browsable pages (auth-walled, JS-heavy)Pricing pages, product tours, profiles
agent-browser
via the 
compound-engineering:agent-browser
skill
MemoryPrior research / decisions / context the user already capturedgrep 
~/.claude/memory/
NotionIf the topic touches a known Notion workspaceDirect Notion API (key in 
$NOTION_API_KEY
, see 
reference_notion_api.md
)
Research archivePrior 
/deep-research
runs that touched this topic		grep 
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/deep-research/archive/
Run discovery passes in parallel where possible. Sequential only when one source needs another's output (e.g., agent-browser a URL discovered by WebSearch).
根据问题类型从以下选项中挑选，并说明选择的来源及理由。
来源适用场景工具
网页搜索（Google）获取权威文章、文档、官方声明
WebSearch
/last30days
了解当前用户真实言论——涵盖Reddit、X、YouTube、HN及近期网页内容
Skill({skill: "last30days", args: "<topic>"})
指定URL用户提供起始URL时
WebFetch
可浏览页面（需认证、JS渲染）定价页、产品演示、个人资料通过
compound-engineering:agent-browser
skill调用
agent-browser
Memory用户已记录的过往研究/决策/上下文grep 
~/.claude/memory/
Notion研究主题涉及已知Notion工作区时直接调用Notion API（密钥存储于
$NOTION_API_KEY
，详见
reference_notion_api.md
）
研究归档过往涉及该主题的
/deep-research
运行记录		grep 
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/deep-research/archive/
尽可能并行执行探索流程。仅当某一来源需要另一来源的输出结果时（例如用agent-browser访问WebSearch发现的URL），才采用串行方式。

Step 3 — Execute discovery

步骤3 — 执行探索

Run each chosen source. For each result, capture:
  • The source (URL or system)
  • 1–3 sentence summary of what was said
  • Date / recency
  • Confidence in the source (high/medium/low)
Don't synthesize yet — just collect.
运行选定的每个来源。针对每个结果，记录：
  • 来源（URL或系统）
  • 1-3句话的内容摘要
  • 日期/时效性
  • 对来源可信度的评估（高/中/低）
暂不进行内容整合——仅做收集工作。

Step 4 — Synthesize

步骤4 — 内容整合

  1. Group findings by theme or sub-question
  2. Contradiction check — flag anywhere sources disagree. Don't average them; surface the disagreement.
  3. Confidence: high (multiple independent sources agree), medium (one strong source or several weak), low (single anecdote or speculation)
  4. Gaps: what would change the answer? What's NOT in the corpus?
  1. 按主题或子问题分组整理研究发现
  2. 矛盾点检查——标注来源存在分歧的地方。不要折中处理，需明确呈现分歧内容。
  3. 可信度评估：高（多个独立来源观点一致）、中（一个权威来源或多个非权威来源）、低（单一轶事或推测）
  4. 信息缺口：哪些内容会改变结论？研究 corpus 中缺少什么？

Step 5 — Output the brief

步骤5 — 输出研究简报

Use this template:
markdown
undefined
使用以下模板：
markdown
undefined

Research: <question>

研究：<问题>

Date: <YYYY-MM-DD> Decision this informs: <one line> Confidence overall: high / medium / low
日期: <YYYY-MM-DD> 决策依据: <一行内容> 整体可信度: 高 / 中 / 低

TL;DR

摘要

<2–4 sentences with the answer>
<2-4句话给出核心结论>

Key findings

关键发现

1. <Finding>

1. <发现内容>

<2–4 sentences>. Sources: [1], [3], [5]
<2-4句话>。来源：[1], [3], [5]

2. <Finding>

2. <发现内容>

...
...

Contradictions / uncertainty

矛盾点/不确定性

  • <where sources disagree, with each side cited>
  • <来源分歧的具体内容及双方引用>

Gaps

信息缺口

  • <what's missing from the corpus>
  • <what to research next to close the gap>
  • <研究 corpus 缺失的内容>
  • <为填补缺口需开展的后续研究方向>

Recommended next steps

建议后续步骤

  1. <action>
  2. <action>
  1. <行动项>
  2. <行动项>

Sources

来源列表

[1] <Title> — <URL or system> (<date>) — <confidence> [2] ...
undefined
[1] <标题> — <URL或系统> (<日期>) — <可信度> [2] ...
undefined

Step 6 — Archive

步骤6 — 归档

Archives live in 
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/deep-research/archive/
(create the directory if missing). Never write archives inside the skill's own folder — skill installs and upgrades re-sync from source and wipe anything saved there. Migration: if this skill's folder contains an old 
references/research-archive/
with user entries, move those files into the archive directory first.
Write the brief to 
<archive dir>/<YYYY-MM-DD>-<slug>.md
so it's grep-able forever. Slug = kebab-case of the topic.
Also append a one-line entry to 
<archive dir>/INDEX.md
(create if missing):
markdown
- 2026-06-15 — [<topic>](./<filename>.md) — <one-line TL;DR>
归档文件存储于
${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/deep-research/archive/
目录（若目录不存在则创建）。切勿将归档文件写入skill自身文件夹——skill的安装与升级会从源端同步并清除该文件夹内的所有本地存储内容。迁移说明：若该skill文件夹内包含旧的
references/research-archive/
目录及用户研究记录，请先将这些文件移至上述归档目录。
将研究简报写入
<归档目录>/<YYYY-MM-DD>-<slug>.md
，以便永久可被grep检索。Slug为主题的短横线分隔式命名（kebab-case）。
同时在
<归档目录>/INDEX.md
文件中添加一行记录（若文件不存在则创建）：
markdown
- 2026-06-15 — [<主题>](./<文件名>.md) — <一行摘要>

Step 7 — Surface

步骤7 — 结果呈现

After archiving:
  • Show the full brief in chat
  • Tell the user the archive path
  • Offer: "Push to Notion or save to a project's docs?"
完成归档后：
  • 在聊天窗口展示完整研究简报
  • 告知用户归档路径
  • 询问用户："是否需要推送到Notion或保存至项目文档？"

Composes with

可组合使用的工具

  • business-brainstorm
    — calls this skill during the market validation step
  • /domain
    — when research includes "is the .com available"
  • /last30days
    — one of the data sources
  • business-brainstorm
    — 在市场验证步骤中调用本skill
  • /domain
    — 当研究涉及“xxx.com域名是否可用”时
  • /last30days
    — 核心数据源之一

Notes on quality

质量注意事项

  • Always cite. Every claim in the brief needs a source pointer.
  • Recency matters — note dates on each source. For fast-moving topics (AI, startups), de-weight sources >12 months old.
  • Don't trust a single source for high-stakes claims. Re-search until you have at least 2 independent corroborations or surface the uncertainty.
  • No padding. If the answer is one paragraph, return one paragraph. The template is a maximum, not a minimum.
  • 务必引用来源：简报中的每一项结论都需要标注来源指向。
  • 时效性至关重要：标注每个来源的日期。对于快速变化的领域（如AI、初创企业），权重降低12个月以上的来源。
  • 高风险结论勿轻信单一来源：需重复研究直至获得至少2个独立佐证，或明确呈现不确定性。
  • 避免冗余内容：若答案仅需一段文字，就只返回一段。模板是上限要求，而非最低标准。