Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计通过

docs文档

Agent Skill

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。

总安装

447

周安装

19

GitHub Stars

公开资料未说明

下载量

157
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:docs(文档)
来源仓库:https://github.com/donutdaniel/agent-skills
仓库路径:skills/docs
安装命令:
npx skills add https://github.com/donutdaniel/agent-skills --skill docs
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/donutdaniel/agent-skills --skill docs

简介

docs 管理项目文档体系,支持同步代码结构、研究规划与决策日志。

  • 适用于维护 README、路线图与决策记录,保持知识资产一致性。
  • 通过 /docs sync、research、retro 命令更新或生成结构化文档。
  • 写入参考文件前需确认路径权限,避免覆盖关键配置或敏感信息。
  • docs 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Docs

Project memory system. Pick an action:

  • /docs sync — Map codebase to references, update roadmap statuses. Use /docs sync structure to only rebuild the folder layout.
  • /docs research — Discover and formalize what to build next into a roadmap doc
  • /docs retro — Analyze logs and decisions over a time period for patterns, wins, and recurring issues

Core Model

  • .docs/reference/ = what exists in the codebase. Sync writes here.
  • .docs/roadmap/ = what should exist but doesn't yet. Research writes here. Organized by status.
  • .docs/decisions/ = decision records. Any command can create these.
  • .docs/logs/ = chronological record of all agent work and decisions. Any command writes here after completing substantial work. Exists so every dev in a shared repo can see what agents did and why.
  • .docs/retros/ = retrospective analyses. Retro writes here. Separate from logs so analytical summaries don't get buried in work history.

Rule Levels

  • Required — Must be followed.
  • Recommended — Strong default guidance.

Global Rules (required)

  1. Command contract — only sync, research, retro.
  2. Docs root + folder contract — this skill manages .docs/ with only these top-level folders:

- .docs/reference/ - .docs/roadmap/ - .docs/decisions/ - .docs/logs/ - .docs/retros/ - Inside .docs/reference/, use: - features/ - architecture/ - conventions/ - runbooks/ - pitfalls/ - Inside .docs/roadmap/, use status folders: - proposed/ — ideas formalized, not started - in-progress/ — actively being worked on - completed/ — fully shipped - archived/ — superseded, rejected, or deprioritized

  1. Naming + date rules

- reference/*kebab-case.md - roadmap/**, decisions/YYYY-MM-DD-kebab-case.md - logs/YYYY-MM-DD-kebab-task-name.md — one log per day per scope of work. Same-day iterations on the same work append as sections; a new day always starts a new file. - Dates must be zero-padded ISO 8601: YYYY-MM-DD.

  1. Append-only history — for decisions/ and retros/, add new files instead of rewriting history. For logs/, same-day iterations on the same work append sections to the existing file (see rule 5).
  2. Log after every complete piece of work — whenever an agent finishes substantial work (pushing a PR, completing a sync, finalizing a research recommendation, shipping a feature), log what was done. One log file per day per scope of work:

- Search today's files in .docs/logs/ (matching today's YYYY-MM-DD prefix) for an existing log matching this work. - If a same-day log exists, append a new --- divider and section to it. - If no same-day log exists, create .docs/logs/YYYY-MM-DD-kebab-task-name.md. - Never append to a log from a previous day — always create a new file. - Each section captures what was done, decisions made, and rationale — so every developer can reconstruct what agents did and why.

  1. Roadmap status lifecycle — track status by moving docs between roadmap folders:

- proposed/in-progress/completed/ - Items that won't be built go to archived/ with a note explaining why (superseded, rejected, deprioritized) - Every roadmap doc must still have a ## Status line for quick context (e.g., Proposed, In Progress, Completed (YYYY-MM-DD), Archived — superseded by [link])

  1. Policy files — if AGENTS.md and/or CLAUDE.md exist, keep their docs-memory guidance aligned.
  2. Path containment — all file operations must target paths within the project root. Reject any path containing .. segments or symlinks that resolve outside the project. All .docs/ operations must stay within .docs/.
  3. No secrets in docs — never write credentials, API keys, tokens, connection strings, or other secrets into any .docs/ file. If sensitive values are encountered during scanning, reference them generically (e.g., "uses environment variable for DB auth") without including the actual value.
  4. Conflict behavior — if repository conventions conflict with these defaults, do not rewrite conventions silently. Report the mismatch and ask the user before changing direction.

Global Rules (recommended)

  • Prefer focused, single-purpose docs.
  • Prefer updating existing docs before creating new ones.
  • Keep docs concrete and scannable.
  • Treat code as ground truth; avoid full schema dumps in docs.

Command References

Each command's full specification lives in its own file under references/:

Match the first argument to the corresponding reference file and follow its instructions.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.74%
按下载量换算55

Claude

28.12%
按下载量换算44

Cursor

19.38%
按下载量换算30

Gemini CLI

9.09%
按下载量换算14

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills