Token导航 LogoToken导航

Agent Skill · 研究检索

doc-coauthoring(文档共同创作)

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。

doc-coauthoring 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,权限为只读;许可证为 MIT,暂无第三方扫描结果,按安装量在研究检索类的 73,675 个 Skill 里排第 64,379。

总安装 303周安装 13GitHub Stars 公开资料未说明下载 106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
MIT

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:doc-coauthoring(文档共同创作)
来源仓库:https://github.com/5dlabs/cto
仓库路径:skills/doc-coauthoring
安装命令:
npx skills add 5dlabs/cto --skill "doc-coauthoring"
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。

AgentSkills.tonpx skills
npx skills add 5dlabs/cto --skill "doc-coauthoring"

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。

doc-coauthoring 用于文档协同编写与结构优化。适用于 README 完善、API 文档生成或用户指南撰写场景。通过 npx skills add 5dlabs/cto --skill "doc-coauthoring" 安装,建议保留原始事实不虚构内容。使用前需统一术语表和风格指南。

应避免过度营销表述保持专业克制。

来源仓库内路径:skills/doc-coauthoring

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 5dlabs/cto --skill "doc-coauthoring";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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

doc-coauthoring 怎么安装?

在终端执行 npx skills add 5dlabs/cto --skill "doc-coauthoring"。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

doc-coauthoring 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。

doc-coauthoring 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

doc-coauthoring 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

doc-coauthoring 用的人多吗?

公开数据是累计安装 303 次,近一周安装 13 次,下载 106 次。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。

doc-coauthoring 安全吗?

没有查到第三方扫描结果,安全性无法从公开资料判断,安装前请自行审查源码、依赖和命令行为。

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。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 没有第三方扫描结果,安全性只能靠自己读源码判断。
  • 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。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 要求有安全扫描结论才肯安装的团队。
  • 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
doc-coauthoring

文档共同创作 · 本页

303—只读MIT未收录

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

Document Co-Authoring Workflow

Guide users through collaborative document creation using three stages: Context Gathering, Refinement & Structure, and Reader Testing.

When to Offer This Workflow

Trigger conditions:

  • Writing documentation, proposals, specs, decision docs, RFCs
  • Substantial writing tasks (not quick notes)
  • Content that will be read by others

Offer the workflow upfront: Explain the three stages and ask if they want this structured approach or prefer freeform.

Stage 1: Context Gathering

Goal: Close the gap between what the user knows and what you know.

Initial Questions

Ask for meta-context (user can answer in shorthand):

  1. What type of document is this? (tech spec, decision doc, proposal)
  2. Who's the primary audience?
  3. What's the desired impact when someone reads this?
  4. Is there a template or format to follow?
  5. Any other constraints or context?

Info Dumping

Encourage the user to dump all context they have:

  • Background on the project/problem
  • Related discussions or documents
  • Why alternatives aren't being used
  • Organizational context
  • Timeline pressures
  • Technical architecture

Tell them: "Don't worry about organizing it - just get it all out."

Clarifying Questions

After initial dump, generate 5-10 numbered questions based on gaps:

1. What's the timeline for this decision?
2. Who are the key stakeholders who need to approve?
3. What happened when you tried approach X?
...

User can answer in shorthand: "1: end of Q1, 2: eng leads + PM, 3: see #channel-name"

Exit condition: Questions show understanding - you can ask about edge cases and trade-offs without needing basics explained.

Stage 2: Refinement & Structure

Goal: Build the document section by section through brainstorming, curation, and iterative refinement.

Process for Each Section

  1. Clarifying Questions - Ask 5-10 questions about what should be included
  2. Brainstorming - Generate 5-20 numbered options based on section complexity
  3. Curation - User indicates what to keep/remove/combine:

- "Keep 1,4,7,9" - "Remove 3 (duplicates 1)" - "Combine 11 and 12"

  1. Gap Check - Ask if anything important is missing
  2. Drafting - Write the section based on selections
  3. Iteration - Refine through surgical edits until satisfied

Section Ordering

Start with whichever section has the most unknowns:

  • Decision docs: Usually the core proposal
  • Specs: Usually the technical approach
  • PRDs: Usually the problem statement

Leave summary sections for last.

Key Instruction for Users

Instead of editing the doc directly, have them indicate what to change:

  • "Remove the X bullet - already covered by Y"
  • "Make the third paragraph more concise"
  • "Move section 3 before section 2"

This helps learn their style for future sections.

Quality Checking

After 3 consecutive iterations with no substantial changes, ask:

"Can anything be removed without losing important information?"

Stage 3: Reader Testing

Goal: Test the document with a fresh perspective to catch blind spots.

Step 1: Predict Reader Questions

Generate 5-10 questions readers might ask when discovering this document:

  • What would they type into search?
  • What would they ask Claude.ai?

Step 2: Test with Fresh Context

If sub-agents available: Invoke a sub-agent with just the document content and each question. Summarize what it got right/wrong.

If no sub-agents: Have user open fresh Claude conversation, paste document, ask the predicted questions. Report back what Reader Claude struggled with.

Step 3: Additional Checks

Ask (or have Reader Claude check):

  • "What in this doc might be ambiguous to readers?"
  • "What knowledge does this doc assume readers already have?"
  • "Are there internal contradictions or inconsistencies?"

Step 4: Fix Blind Spots

For each issue found, loop back to Stage 2 refinement for that section.

Exit condition: Reader Claude consistently answers questions correctly and doesn't surface new gaps.

Final Review

When Reader Testing passes:

  1. Recommend they do a final read-through themselves
  2. Suggest double-checking facts, links, technical details
  3. Ask them to verify it achieves the intended impact

Final tips:

  • Consider linking this conversation in an appendix
  • Use appendices for depth without bloating main doc
  • Update as feedback comes from real readers

Handling Deviations

SituationResponse
User wants to skip a stageAsk if they want to skip and write freeform
User seems frustratedAcknowledge time investment, suggest faster path
Missing context on something mentionedAsk proactively, don't let gaps accumulate
User edits doc directlyNote changes, incorporate preferences for future sections

Tips for Effectiveness

  • Be direct and procedural
  • Explain rationale briefly when it affects behavior
  • Don't try to "sell" the approach - just execute it
  • Give user agency to adjust the process
  • Quality over speed - each iteration should make meaningful improvements

平台分布

Claude Code

29.61%
按下载量换算31

windsurf

21.48%
按下载量换算23

trae

19.22%
按下载量换算20

OpenCode

11.88%
按下载量换算13

Codex

8.09%
按下载量换算9

Antigravity

3.32%
按下载量换算4

安全审计

暂无安全审计结果可展示。

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。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