Token导航 LogoToken导航TokenDH.com
待分类权限需确认github未标认证来源可访问许可证需确认审计异常

tiangong-wiki-skill天工维基技能

Agent Skill

tiangong-wiki-skill 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

184

周安装

12

GitHub Stars

4

下载量

97
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:tiangong-wiki-skill(天工维基技能)
来源仓库:https://github.com/biaoo/tiangong-wiki
仓库路径:skills/tiangong-wiki-skill
安装命令:
npx skills add https://github.com/biaoo/tiangong-wiki --skill tiangong-wiki-skill
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/biaoo/tiangong-wiki --skill tiangong-wiki-skill

简介

用于处理 GitHub 仓库、Issue 和 Pull Request 协作信息。

  • 适合围绕代码变更、仓库状态进行整理和分析。
  • 可结合项目上下文理解代码修改意图和影响范围。
  • 安装命令:npx skills add https://github.com/biaoo/tiangong-wiki --skill tiangong-wiki-skill
  • 安装前建议确认权限范围和维护状态。

SKILL.md

Wiki Skill

Core Goal

Use the local wiki as the durable knowledge layer — not just for the current conversation, but for all future work. Query first, then read or edit the Markdown files that remain the source of truth.

CLI Entrypoint

Use tiangong-wiki <command> on macOS, Linux, WSL, and Git Bash.

On Windows native shells such as PowerShell, Command Prompt, background daemon tasks, or Codex worker automation, use the npm command shim explicitly:

tiangong-wiki.cmd doctor
tiangong-wiki.cmd sync
tiangong-wiki.cmd lint --format json

Do not invoke the suffixless tiangong-wiki executable from Windows native shells. npm also installs that shebang script for POSIX-like environments, but Windows may treat it as an unknown file and open the "choose an app" dialog instead of executing the CLI.

When to Use

Activate this skill in three scenarios:

1. Knowledge Retrieval

You need historical knowledge, methods, behavioral patterns, or prior decisions before answering or acting.

Trigger: The current task could benefit from previously captured insights, or the user asks about something that may already exist in the wiki.

Minimal steps:

  1. Choose a retrieval strategy based on what you know:

- Know the type, tag, or metadata → tiangong-wiki find - Have a keyword or short phrase → tiangong-wiki fts - Fuzzy intent, embeddings configured → tiangong-wiki search - Need graph context or related pages → tiangong-wiki graph

  1. Inspect the best candidate with tiangong-wiki page-info <id>.
  2. Read the Markdown file if you need full content.

Stop when: you have enough context to proceed, or no relevant pages exist.

2. Knowledge Capture

A conversation, task, or other skill produced a high-value insight, method, correction, or piece of information worth preserving.

Trigger: You recognize durable knowledge — something that would be useful to retrieve in a future context.

Two paths:

  • Direct page creation — when you can clearly identify the page type and structure the content yourself:

1. Search for existing pages to avoid duplicates (find, fts, or search). 2. Discover the ontology: tiangong-wiki type recommend --text "<summary>" --keywords "a,b,c" --format json. 3. Update an existing page if one covers the same knowledge object. Create a new page only when the object is distinct. 4. After writing: tiangong-wiki sync --path <page-id> then tiangong-wiki lint --path <page-id> --format json.

  • Save to vault — when the material is raw, complex, or needs further processing:

1. Save the file to vault/. 2. The vault-to-wiki workflow will handle triage and page creation. 3. See references/vault-to-wiki-instruction.md for the full decision model.

Key rules:

  • Do not assume any page type is the default destination. Always discover via CLI.
  • Prefer updating the best existing page over creating near-duplicates.
  • Only use frontmatter fields declared by the chosen type (tiangong-wiki type show <type>). Do not invent ad-hoc fields.
  • Preserve provenance with sourceRefs and type-specific source fields.

3. Knowledge Maintenance

You discover that existing wiki content is outdated, incorrect, or incomplete during retrieval or conversation.

Trigger: A retrieved page contains stale information, or new evidence contradicts existing content.

Minimal steps:

  1. Locate the page: tiangong-wiki find or tiangong-wiki page-info <id>.
  2. Edit the Markdown file — keep edits minimal and provenance-preserving.
  3. After editing: tiangong-wiki sync --path <page-id> then tiangong-wiki lint --path <page-id> --format json.

For systematic maintenance (health checks, orphan cleanup, stale content review), see references/wiki-maintenance-instruction.md.

References (read on demand)

DocumentRead when...
references/cli-interface.mdYou need full command options, flags, or output format details
references/vault-to-wiki-instruction.mdProcessing vault files into wiki pages
references/wiki-maintenance-instruction.mdRunning systematic maintenance or health checks
references/template-design-guide.mdDesigning a new page type or evolving an existing template
references/troubleshooting.mdEnvironment setup, configuration issues, or error diagnosis

Assets

  • assets/wiki.config.default.json — default configuration template
  • assets/templates/ — Markdown page templates for each registered type

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

34.78%
按下载量换算34

Claude

30.68%
按下载量换算30

Cursor

20.23%
按下载量换算20

Gemini CLI

10.96%
按下载量换算11

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

可疑

权限和风险

权限需确认

当前来源未能明确判断权限范围,默认进入异常复核队列。

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 warning/failed 结果,不能写成本站确认安全。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