Token导航 LogoToken导航

Agent Skill · 研究检索

living-docs(生活文档)

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。

living-docs 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,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 65,896。

总安装 272周安装 11GitHub Stars 公开资料未说明下载 85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:living-docs(生活文档)
来源仓库:https://github.com/rcrespodev/rcrespodev-skills
仓库路径:skills/living-docs
安装命令:
npx skills add https://github.com/rcrespodev/rcrespodev-skills --skill living-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/rcrespodev/rcrespodev-skills --skill living-docs

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。

用于文档、README 和内容稿件的整理与改写,适合提炼结构与统一术语。可补齐章节、检查链接或整合零散材料为可读文档。使用时应保留项目事实与路径,避免将未确认信息写成确定结论。安装方式:通过 npx 从 GitHub 仓库添加,需确认权限与维护状态。注意:对外文案需控制语气,避免过度营销或夸大能力。

来源仓库内路径:skills/living-docs

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/rcrespodev/rcrespodev-skills --skill living-do…;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

living-docs 怎么安装?

在终端执行 npx skills add https://github.com/rcrespodev/rcrespodev-skills --skill living-docs。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

living-docs 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。

living-docs 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。

living-docs 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。

living-docs 用的人多吗?

公开数据是累计安装 272 次,近一周安装 11 次,下载 85 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。

living-docs 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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
living-docs

生活文档 · 本页

272—需要联网未明确未通过

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

Living Docs

Generate documentation driven by actual code changes. Every document traces to specific commits and files. When code changes, docs change.

Language policy: Generate all documentation in English. Include Spanish translations in the aliases frontmatter field.

Folder Structure

All generated docs live under ${GIT_REPO_ROOT}/docs/ (relative to the repository root). Create this structure if it doesn't exist:

docs/
├── components/          # Component Docs (services, apps, libraries)
│   ├── auth-service.md
│   ├── user-service.md
│   └── shared-utils.md
├── changelogs/          # Changelogs and release notes
│   ├── changelog-auth-service-2026-02-13.md
│   └── changelog-global-2026-02-13.md
├── adrs/                # Architecture Decision Records
│   ├── adr-001-redis-caching.md
│   └── adr-002-event-driven-auth.md
├── runbooks/            # Runbooks and SOPs
│   ├── runbook-deploy-auth-service.md
│   └── runbook-database-migration.md
└── index.md             # Auto-generated index linking all docs

File Naming Conventions

Doc TypeDirectoryFilename Pattern
Component Docdocs/components/{component-name}.md
Changelogdocs/changelogs/changelog-{scope}-{YYYY-MM-DD}.md
ADRdocs/adrs/adr-{NNN}-{slug}.md
Runbookdocs/runbooks/runbook-{operation-slug}.md
Indexdocs/index.md

Index File

After generating or updating docs, update docs/index.md with links to all docs:

# Documentation Index

> Auto-generated. Last updated: YYYY-MM-DD

## Components
- [[auth-service]] — Authentication and authorization service
- [[shared-utils]] — Shared utility library

## Recent Changelogs
- [[changelog-auth-service-2026-02-13]] — Added OAuth2 support

## Architecture Decisions
- [[adr-001-redis-caching]] — Accepted

## Runbooks
- [[runbook-deploy-auth-service]] — Deployment procedure

Workflow

1. Gather context     → Determine diff scope
2. Extract diff data  → Run extract-diff.sh or git commands
3. Classify changes   → Identify what matters (see references/analysis-patterns.md)
4. Select templates   → Pick doc types (see references/templates.md)
5. Generate docs      → Write markdown with full YAML frontmatter
6. Verify output      → Cross-check generated docs against diff
7. Present summary    → Show what was generated and why

Step 1: Gather Context

Determine from the user's message (ask only if not inferrable):

  • Diff scope: Branch comparison (e.g., feature/X vs Develop) or last N commits?
  • Existing docs: Any docs to update rather than create from scratch?

Defaults (when not specified):

  • Diff scope: Current branch vs default branch (auto-detected from repo)
  • Output path: Always ${GIT_REPO_ROOT}/docs/ (using the folder structure above)
  • Existing docs: Search ${GIT_REPO_ROOT}/docs/ for matching filenames before creating new docs

Step 2: Extract Diff Data

Run the extraction script from this skill's directory:

# Branch comparison (auto-detects default branch)
bash <this-skill-path>/scripts/extract-diff.sh <repo-path> --branch <target>

# Last N commits
bash <this-skill-path>/scripts/extract-diff.sh <repo-path> --commits 20

# Filtered by path (for monorepos)
bash <this-skill-path>/scripts/extract-diff.sh <repo-path> --branch Develop --path services/auth/

For multi-repo workspaces, run per repository.

Large Diff Strategy

When the diff is large (100+ files changed or output exceeds ~50KB):

  1. Start with --stat only — Use the file list and change counts to plan
  2. Read full diff only for high-impact files — APIs, schemas, configs, contracts, new files
  3. Read only changed hunks for medium-impact files — Business logic, services
  4. Skip full diff for low-impact files — Tests, formatting, comments
  5. Split by directory — If still too large, analyze one component/service at a time

Step 3: Classify Changes

Read references/analysis-patterns.md for the full classification guide.

Priority: High-impact changes (new components, public API changes, business logic changes, schema migrations, new dependencies, infra changes) are always documented. Medium-impact (internal refactors, test changes) only if significant. Low-impact (formatting, comments, patch bumps) are skipped.

Key distinctions (from analysis-patterns.md):

  • Business logic vs refactor: Did test expectations change? → business logic. Same tests pass? → refactor.
  • Interface exposure: External (public API) > Inter-component (shared packages) > Internal (same module). Document proportionally.

Map each significant change to doc types:

Change TypeImpactDoc Types
New component / moduleHighComponent Doc (new)
Public API changeHighComponent Doc + Changelog (Breaking if contract changed)
Business logic changeHighComponent Doc + Changelog
Inter-component interface changeHighComponent Doc + flag downstream consumers
Schema / data model changeHighADR + Component Doc
Major dependency addedMedium-HighADR + Component Doc
Infrastructure / deployment changeMediumRunbook
Internal refactor (same behavior)LowChangelog (Internal) or skip
Release milestone—Changelog

Step 4: Select and Fill Templates

Read references/templates.md for all templates and the frontmatter schema.

Frontmatter rules:

  • ALWAYS include all required fields: aliases, type, layer, status, owner, tech_stack, last_updated, source_branch, commit_range
  • Use [[wiki-links]] for owner, tech_stack, and cross-references
  • Set last_updated to today's date
  • Set source_branch and commit_range from the actual diff
  • Set status honestly: active, debt, zombie, or gap
  • Populate aliases with English keywords + Spanish equivalents
  • Infer owner and tech_stack using the heuristics in templates.md → "Inferring Frontmatter from Diffs"
  • Omit template sections that don't apply (e.g., don't include "Events Published" for a REST-only service)

Step 5: Generate Docs

Write docs to the folder structure defined above.

Updating existing docs: Read first, preserve frontmatter structure, update last_updated and commit_range, modify only affected sections, append to "Recent Changes".

Merge Strategy for Incremental Updates

When a doc already exists:

SectionStrategy
FrontmatterMerge: Update last_updated, commit_range, status. Preserve owner, aliases (append new ones)
What It DoesReplace only if the component's purpose fundamentally changed
API Surface / Exported APIMerge: Add new entries, update changed entries, mark removed entries as deprecated
DependenciesReplace with current state
ConfigurationMerge: Add new env vars, update changed ones
Key FilesReplace with current state
Recent ChangesAppend new changes at the top, keep last 5-10 entries

Step 6: Verify Output

Before presenting to the user, cross-check:

  • Every endpoint/export mentioned in docs exists in the diff or codebase
  • All frontmatter required fields are populated (no empty or placeholder values except owner: "[[TBD]]")
  • File paths referenced in "Key Files" actually exist
  • Breaking changes flagged in Changelog match actual contract changes in the diff
  • No duplicate docs (check existing files before creating new ones)

Step 7: Present Summary

Show the user what was generated:

## Documentation Generated

| File | Type | Reason |
|------|------|--------|
| docs/components/auth-service.md | Component Doc | New endpoints in routes/users.ts |
| docs/changelogs/changelog-auth-2026-02-13.md | Changelog | 12 commits with 3 features, 2 fixes |

### Key Changes Documented
- [bullets]

### Skipped (Low Impact)
- [what and why]

Always update ${GIT_REPO_ROOT}/docs/index.md with links to all generated docs (for multi-repo workspaces, do this per repository).


Quality Rules

  • No fluff: Every sentence carries information. Cut filler.
  • Trace to code: Every claim references a file, commit, or config.
  • Tables over prose: For endpoints, env vars, dependencies — always tables.
  • Be honest: If the diff reveals tech debt, set status: debt. Living docs tell the truth.
  • Aliases matter: Include concept name, Spanish translation, common abbreviations.
  • Omit empty sections: Don't include template sections that have no content for this component.

Few-Shot Example

Input: Diff Summary

=== FILE STATS ===
 services/auth/src/routes/auth.ts   | 45 +++++++++--
 services/auth/src/services/oauth.ts | 120 ++++++++++++++++++++++++++++
 services/auth/src/types/auth.dto.ts |  15 ++++
 packages/shared-types/src/user.ts   |   8 ++--
 services/auth/package.json          |   2 +  (added passport-google-oauth20)
 services/auth/tests/oauth.test.ts   |  85 ++++++++++++++++++++

=== COMMIT LOG ===
a1b2c3d feat: add Google OAuth2 login flow
d4e5f6g feat: add OAuth callback handler
h7i8j9k fix: handle missing email in OAuth profile
l0m1n2o chore: add passport-google-oauth20 dependency

Output: Classification

  1. New OAuth service file (oauth.ts, 120 lines) → High impact, new feature → Component Doc update
  2. Route changes (auth.ts, 45 lines) → Public API change → Component Doc + Changelog
  3. Shared types change (packages/shared-types/user.ts) → Inter-component interface → Component Doc + flag consumers
  4. New major dependency (passport-google-oauth20) → ADR candidate
  5. New tests with different expectations → Confirms business logic change, not refactor

Output: Generated Docs

  • docs/components/auth-service.md — Updated: added OAuth endpoints, new dependency, new key files
  • docs/changelogs/changelog-auth-2026-02-13.md — New: 2 features (OAuth login, callback handler), 1 fix
  • docs/adrs/adr-003-google-oauth.md — New: Decision to use passport-google-oauth20 for social login
  • docs/components/shared-types.md — Updated: flag that UserDTO interface changed (consumed by 2 services)

Resources

  • scripts/extract-diff.sh — Extract structured diff data (file stats, commit log, full diff)
  • references/templates.md — All doc templates with frontmatter schema
  • references/analysis-patterns.md — How to classify changes from diffs

平台分布

Codex

35.49%
按下载量换算30

Claude

31.84%
按下载量换算27

Cursor

18.55%
按下载量换算16

Gemini CLI

9.08%
按下载量换算8

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

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