Token导航 LogoToken导航

Agent Skill · 研究检索

content-research-writer(内容研究作家)

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

content-research-writer 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 47,767。

总安装 720周安装 30GitHub Stars 1下载 240核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:content-research-writer(内容研究作家)
来源仓库:https://github.com/pixel-process-ug/superkit-agents
仓库路径:skills/content-research-writer
安装命令:
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill content-research-writer
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill content-research-writer

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

content-research-writer 用于辅助文档、README 和内容稿件的整理与改写,适合结构提炼和术语统一。适用于 Markdown、说明文和外文案的章节补齐和可读性提升场景。保留项目已有事实和路径,避免将未确认信息写成确定结论。

通过 npx skills add 命令从指定 GitHub 仓库安装并使用该技能。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/content-research-writer

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill conte…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

content-research-writer 怎么安装?

在终端执行 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill content-researc…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

content-research-writer 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

content-research-writer 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

content-research-writer 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

content-research-writer 用的人多吗?

公开数据是累计安装 720 次,近一周安装 30 次,下载 240 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

content-research-writer 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
content-research-writer

内容研究作家 · 本页

7201需要联网未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Content Research Writer

Overview

Produce well-researched, authoritative long-form content with proper source attribution, structured argumentation, and evidence-based claims. This skill covers research methodology, source evaluation, citation management, outline construction, drafting workflows, fact-checking protocols, and publishing-ready formatting for articles, whitepapers, reports, and educational content.

Apply this skill whenever content must be backed by evidence, cited properly, and structured for credibility with expert audiences.

Multi-Phase Process

Phase 1: Research Planning

  1. Define the topic scope and target audience
  2. Identify key questions the content must answer
  3. Determine content type (article, whitepaper, case study, guide)
  4. Set word count target and depth level
  5. Establish credibility requirements (peer-reviewed, industry reports, primary data)
  6. Create a research timeline with milestones
STOP — Do NOT begin source discovery until the research plan is documented and scope is agreed upon.

Phase 2: Source Discovery and Evaluation

  1. Search academic databases (Google Scholar, PubMed, JSTOR, arXiv)
  2. Identify industry reports and authoritative publications
  3. Find primary sources (official documentation, datasets, specifications)
  4. Evaluate source credibility using the CRAAP test (see table below)
  5. Organize sources in a reference manager or structured format
  6. Extract key findings, statistics, and quotable passages
STOP — Do NOT begin outlining until you have sufficient Tier 1-2 sources for every core claim.

Phase 3: Outline and Structure

  1. Create a thesis statement or central argument
  2. Build hierarchical outline with main sections and subsections
  3. Map evidence to each section (which sources support which claims)
  4. Identify gaps requiring additional research
  5. Define transitions between sections for narrative flow
  6. Plan visual elements (tables, charts, diagrams, callouts)
STOP — Do NOT begin drafting until the outline is reviewed and evidence gaps are filled.

Phase 4: Drafting

  1. Write section by section following the outline
  2. Integrate citations as you write (never retrofit)
  3. Balance original analysis with supporting evidence
  4. Use topic sentences and clear paragraph structure
  5. Include concrete examples, data points, and case studies
  6. Write introduction last (after body is complete)
STOP — Do NOT move to review until all sections are drafted and all citations are in place.

Phase 5: Review and Fact-Check

  1. Verify every factual claim against its source
  2. Check all statistics for accuracy and context
  3. Ensure citations are complete and correctly formatted
  4. Review for logical consistency and argument strength
  5. Proofread for clarity, grammar, and style
  6. Have subject matter expert review if possible

Content Type Decision Table

Content TypeWord CountResearch DepthAudienceCitation Style
Blog article1,500-3,000Tier 2-3 sourcesGeneralInline links
Long-form article3,000-5,000Tier 1-3 sourcesInformed readersParenthetical (APA)
Whitepaper3,000-8,000Tier 1-2 mandatoryDecision-makersFootnotes or numbered
Case study1,000-2,500Primary + Tier 2BuyersInline attribution
Technical report5,000-15,000Tier 1 mandatoryExpertsIEEE or APA
Educational guide2,000-6,000Tier 1-3 mixedLearnersParenthetical

Source Evaluation Framework (CRAAP Test)

CriterionQuestions to AskRed Flags
CurrencyWhen was it published? Updated?> 5 years old for fast-moving topics
RelevanceDoes it directly address your topic?Tangential connection, different context
AuthorityWho is the author? What are their credentials?No author, no institutional affiliation
AccuracyIs it supported by evidence? Peer-reviewed?No citations, unverifiable claims
PurposeWhy does this exist? Inform, sell, persuade?Strong commercial bias, advocacy without disclosure

Source Tier System

TierSource TypeCredibilityUse For
Tier 1Peer-reviewed journals, official standardsHighestCore claims, statistics
Tier 2Industry reports (Gartner, McKinsey), textbooksHighMarket data, frameworks
Tier 3Reputable news outlets, official documentationGoodContext, current events
Tier 4Expert blogs, conference talks, interviewsModeratePerspectives, opinions
Tier 5Social media, forums, WikipediaLowDiscovery only, never cite directly

Citation Formats

Inline Citation Styles

# Parenthetical (APA-style)
Research shows that 73% of enterprises have adopted cloud-native architectures
(Smith & Johnson, 2025).

# Narrative
According to Smith and Johnson (2025), 73% of enterprises have adopted
cloud-native architectures.

# Footnote-style (Chicago)
Research shows significant cloud adoption.[^1]

[^1]: Smith, J., & Johnson, R. (2025). Cloud adoption trends.
*Journal of Cloud Computing*, 12(3), 45-62.

# Numbered (IEEE-style)
Research shows significant cloud adoption [1].

## References
[1] J. Smith and R. Johnson, "Cloud adoption trends," *J. Cloud Computing*,
    vol. 12, no. 3, pp. 45-62, 2025.

Reference Format Templates

# Journal Article
Author, A. B., & Author, C. D. (Year). Title of article. *Journal Name*,
Volume(Issue), Pages. https://doi.org/xxxxx

# Book
Author, A. B. (Year). *Title of book* (Edition). Publisher.

# Website
Author or Organization. (Year, Month Day). Title of page. Site Name.
https://www.example.com/page

# Report
Organization. (Year). *Title of report* (Report No. XXX).
https://www.example.com/report.pdf

Content Structure Templates

Long-Form Article (2000-5000 words)

1. Hook / Opening Anecdote (100-200 words)
2. Context and Problem Statement (200-300 words)
3. Thesis / Key Insight (50-100 words)
4. Section 1: Background (400-600 words)
   - Historical context
   - Current state
   - Key definitions
5. Section 2: Core Analysis (600-1000 words)
   - Main argument with evidence
   - Data and statistics
   - Expert perspectives
6. Section 3: Implications (400-600 words)
   - Practical applications
   - Case studies
   - Future outlook
7. Section 4: Counterarguments (200-400 words)
   - Acknowledge limitations
   - Address objections
8. Conclusion (200-300 words)
   - Synthesize key findings
   - Call to action or forward-looking statement
9. References

Whitepaper (3000-8000 words)

1. Executive Summary (300-500 words)
2. Introduction and Problem Statement (500-800 words)
3. Methodology (300-500 words)
4. Findings / Analysis (1500-3000 words)
   - Section with data visualization
   - Comparative analysis
   - Case studies
5. Recommendations (500-1000 words)
6. Conclusion (300-500 words)
7. Appendices
8. References

Case Study (1000-2500 words)

1. Executive Summary (100-200 words)
2. Challenge / Problem (200-400 words)
3. Approach / Solution (300-600 words)
4. Implementation (300-500 words)
5. Results and Metrics (200-400 words)
6. Lessons Learned (200-300 words)
7. About [Company/Subject]

Writing Quality Checklist

Paragraph Level

  • Each paragraph has a clear topic sentence
  • Paragraphs are 3-6 sentences (avoid walls of text)
  • Transitions connect paragraphs logically
  • Evidence follows claims immediately

Sentence Level

  • Vary sentence length (mix short and long)
  • Active voice preferred over passive
  • Avoid jargon without definition
  • Concrete language over abstract

Document Level

  • Introduction establishes the "so what" clearly
  • Each section advances the central argument
  • No unsupported claims
  • Conclusion adds value (doesn't just repeat)
  • Consistent tone and reading level throughout

Fact-Checking Protocol

Verification Steps

  1. Primary source check: Trace every claim back to its original source
  2. Cross-reference: Verify key facts with at least 2 independent sources
  3. Statistical validation: Check that numbers are current and in context
  4. Quote accuracy: Verify exact wording of all direct quotes
  5. Date verification: Confirm all dates and timelines
  6. Name and title check: Verify correct spelling and current titles

Common Fact-Checking Pitfalls

PitfallExamplePrevention
Outdated statistics"50% of..." from a 2018 studyAlways note publication year, seek recent data
Misattributed quotesEinstein didn't say most "Einstein quotes"Trace to primary source document
Survivorship bias"All successful companies do X"Look for counterexamples
Correlation as causation"Countries that eat chocolate win more Nobels"Distinguish correlation from causation
Out-of-context numbers"Revenue grew 500%" (from $1 to $5)Always provide absolute numbers and context

Research Tools

ToolPurposeBest For
Google ScholarAcademic paper searchPeer-reviewed research
Semantic ScholarAI-powered paper discoveryFinding related work
arXivPreprintsCutting-edge CS/ML/Physics
PubMedMedical/bio researchHealth and life sciences
StatistaStatistics and market dataIndustry data points
Wayback MachineHistorical web pagesVerifying past claims
Zotero / MendeleyReference managementOrganizing sources
PerplexityAI-assisted researchInitial discovery

Anti-Patterns / Common Mistakes

Anti-PatternWhy It FailsWhat To Do Instead
Claims without citationsUndermines credibility entirelyCite every factual claim inline
Single-source assertionsOne source can be wrong or biasedCross-reference with 2+ independent sources
Citing secondary when primary existsTelephone game distorts findingsTrace to and cite the original study
Writing introduction firstLeads to misalignment with bodyWrite body first, introduction last
Padding to reach word countReaders detect filler immediatelyAdd depth or cut the target
Weasel words without specifics"Some experts say" means nothingName the expert, cite the source
Retrofitting citations after draftingGaps in evidence go unnoticedIntegrate citations during writing
Ignoring counterargumentsOne-sided work lacks credibilityAddress objections explicitly
Paraphrasing too closelyBorderline plagiarism even with citationSummarize in your own analytical voice
No conflict-of-interest disclosureErodes trust when discoveredDisclose sponsorship or affiliations upfront

Anti-Rationalization Guards

  • Do NOT skip the CRAAP test because "the source looks reputable" -- evaluate it formally.
  • Do NOT use Tier 4-5 sources for core claims, regardless of convenience.
  • Do NOT begin drafting without a completed outline with evidence mapped to sections.
  • Do NOT publish without running the fact-checking protocol on every statistic and quote.
  • Do NOT retrofit citations after writing -- integrate them as you draft.

Integration Points

SkillHow It Connects
seo-optimizerResearch content needs SEO-optimized titles, meta descriptions, and structured data
content-creatorResearch findings feed into marketing copy and social media content
email-composerResearch summaries inform stakeholder update emails and executive briefings
tech-docs-generatorTechnical research follows similar source evaluation and citation practices
llm-as-judgeEvaluate research content quality against rubric dimensions
clean-codeWriting quality checklist parallels clean code principles for prose

Skill Type

FLEXIBLE — Adapt research depth, citation formality, and structure to the content type and audience. Academic whitepapers demand Tier 1 sources and formal citations; blog posts may use a lighter approach. The fact-checking protocol and source evaluation framework are always recommended.

平台分布

Codex

34.39%
按下载量换算83

Claude

34.25%
按下载量换算82

Cursor

18.66%
按下载量换算45

Gemini CLI

10.23%
按下载量换算25

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills