Ljg Push logo

Ljg Push

CommunityPopular
lijigang
ljg-push

把 ~/.agents/skills/ljg-* 里所有更新过的 skills 同步到 github repo (ljg-skills),先推 master 分支(org-mode 输出风格),再切 md 分支(markdown 输出风格)做基础 markdown 化后推。Use when user says '/ljg-push', 'push skills', '推送 skills', '同步 skills', 'sync ljg', or whenever ljg-* skills get updated and need shipping. NOT FOR pushing non-ljg skills or arbitrary git repos.

Overview

Publisherlijigang
Repositoryljg-skills
Skill nameljg-push
Stars
7.4K
Forks
834
Bundled files
5
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 5 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by lijigang on GitHub. Read the source before you install it.

Installation

Install the Ljg Push AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/lijigang/ljg-skills.git /tmp/ljg-skills
mkdir -p .claude/skills
cp -r /tmp/ljg-skills/skills/ljg-push .claude/skills/ljg-push
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ljg Push in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Ljg Push on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Ljg Push is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

ljg-push: 推送 ljg-* skills

把本地 ~/.agents/skills/ljg-* 里改过的 skills,一键同步到 github repo,覆盖 master 和 md 两个分支。

仓库路径(硬编码)

SKILLS_REPO="$HOME/code/ljg-skills"     # 本地工作 repo
SKILLS_LOCAL="$HOME/.agents/skills"      # 本地 skill 源
REPO_URL="git@github.com:lijigang/ljg-skills.git"

如果 $SKILLS_REPO 不存在,脚本会自动 clone。如果它存在但不是 ljg-skills 的 git repo,脚本会报错退出(不破坏现有目录)。

两条分支的差异

分支输出格式文件扩展加粗文件头
master(默认)org-mode.org*bold*#+title:
mdmarkdown.md**bold**YAML frontmatter

~/.agents/skills/ 里的 skill 是 master 风格(源版本)。md 分支的差异由脚本自动转换 + 必要时手工补。

脚本推完 md 后会自动切回 master。本地 $HOME/code/ljg-skills 应该始终停在源分支,方便下次查看和安装。

工作流

Workflows/Push.md 步骤执行 → 调用 Tools/Push.sh

README 一致性(硬 gate)

每次 push 前,脚本强制做一件事:把 README 跟 local skills 对一遍

  • 列出 ~/.agents/skills/ljg-* 全部 skill 名
  • grep $SKILLS_REPO/README.md 里出现的 ljg-xxx
  • 找出 local 有但 README 没有的——几乎肯定意味着 README 漏更新
  • 命中 → push 中止,报告差异

每次 push 都是检视 README 的机会。问自己:

  1. 新增 skill 了吗?README 的 skill 清单 / 安装命令需要加一行
  2. 删了 skill 吗?README 对应行要删
  3. 某个 skill 的描述大改了吗?README 的简介可能要同步

确认 README 已审、确实不需要更新时,绕过 gate:

bash
/ljg-push --skip-readme-check

自动转换的范围

md 分支同步时自动转换(2026-06-12 起含 org 文件本体):

Org 文件中的 #+begin_example / #+end_example 图块转为 text 围栏,保留块内文字,不将其中的星号或井号误当标题和注释。标记兼容大小写。

  • org 文件本体:skill 内每个 .org 文件(assets/ 除外)转成同名 .md 并删除原件——org 头块→YAML frontmatter(含 --- 围栏,filetagstags)、* 标题→# 标题(层级保留)、#+ATTR_* 行删除、[[file:x]]![](x)#+begin_src→``` 围栏。Markdown 与运行时文本文件(如 .ts / .js / .json / .sh)里对实际被改名文件的引用同步改写
  • Markdown 内嵌的完整 Org 示例```org 模板,以及首行就是 #+key: 的无语言围栏模板,都会转为 ```markdown;连续头块变为带 --- 的 YAML frontmatter,标题、链接、强调、等宽文本与分隔线随模板一起转换;原文件使用纯 CRLF 时保留其换行风格
  • 文件扩展引用:__qa.org__qa.md__paper.org__paper.md 等(denote 命名约定)
  • 关键词:org-modemarkdownOrg-modeMarkdown
  • org 式格式指令:加粗用 *bold*(单星号)…加粗用 **bold**(双星号)标题层级从 * 开始从 # 开始Org 文件头Markdown 文件头、行首 #+title: 等 8 个示例键 → YAML 键行
  • 结构化强调标签:行首 - *标签*:- **标签**:;紧邻这类标签的 org 围栏同步改为 markdown
  • 输出契约与运行时默认值:evals/*.json 中明确要求保存 Org 的提示与期望改为 Markdown;stdin__*.org 默认值改为 .md,同时保留显式 Org 输入与测试夹具
  • ljg-is 的生成合同:输出标题、默认文件扩展、路径称谓、元数据名、tags 与双格式验收句统一切到 Markdown/YAML + Denote,并删除 Org 专用 lint 阻断表述;提交前残留审计拒绝任何未转换的 Org 输出指令

检测与同步共用同一文件边界:按校验和比较内容,可靠识别文件新增、删除、内容和可执行位变化;纯时间戳、目录元数据、空目录不会单独触发发布,任意层级的 .git/node_modules/.DS_Store 不进入同步。忽略清单只含这三类明确杂物,.bak.backup 等真实备份仍会触发同步和发布审查。

仍不自动转换(按需手工):

  • 正文里的 *bold* 标记:markdown 里 *x* 是斜体,盲替会破坏文档自身格式;只有「行首 bullet + 全角冒号」标签会安全转换

Voice Notification

bash
curl -s -X POST http://localhost:31337/notify \
  -H "Content-Type: application/json" \
  -d '{"message": "Running Push in ljg-push"}' \
  > /dev/null 2>&1 &

输出文本:Running **Push** in **ljg-push**...

Examples

Example 1: 一键推送

User: /ljg-push
→ 检测 ~/.agents/skills/ljg-* 中跟 repo 有差异的 skills
→ master: rsync + bump version + commit + push
→ md: rsync + mdize + bump version + commit + push
→ 切回 master
→ 报告:哪些 skills 推了,新版本号,剩余手工差异

Example 2: 看会推什么但不真推

User: /ljg-push --dry-run
→ 列出会被同步的 skills
→ 列出会做的 markdown 化转换
→ 不执行 rsync / commit / push

Gotchas

  • README 漂移是最容易被忽略的——加完新 skill 直接推,README 还停在老清单。脚本现在有硬 gate 拦这一刀;拦下来时不要无脑加 --skip-readme-check,先去看一下 README
  • 脚本前提是 git credentials 已配好(ssh key 或 PAT)—— ljg-push 不处理认证,认证失败时直接报错
  • master 必须先推——md 分支的 markdown 化基于 master 的 org 版本做转换。反过来推会破坏顺序
  • untracked 杂物(如 assets/measure.js)会被 rsync 同步到 repo——如果不想推,先在本地删掉,或加进 .gitignore
  • 同步忽略项必须窄而一致——.git/node_modules/.DS_Store 不进入同步;纯时间戳、目录元数据和空目录不会单独触发发布,也不会进入 Git 结果。普通备份文件不在忽略清单里,仍会被检测出来,避免把可能公开的真实内容藏掉
  • org 文件本体已自动转换(2026-06-12 起)——template.org 等会被转成 .md 并删除原件,每次推送重新生成(rsync --delete 冲掉也无妨,幂等)。遗留手工项只剩正文里的 *bold* 标记。新增带复杂构件的 org reference 文件后,先 --dry-run 或沙盒跑一遍 mdize 看转换效果
  • 重命名引用可能藏在运行时代码里——例如测试用 new URL("../Template.org", import.meta.url) 读取模板;只改 Markdown 文档会让 md 分支缺文件。转换器会按本次实际转出的 basename 精确改写 .ts / .js / .json / .sh 等文本消费者,同时保留没有对应实体文件的 Org 测试夹具字符串
  • 结构化标签不能靠枚举示例词——xf(x) 之外还会出现「主体/边界」「代入」等真实标签;转换器按 - *标签*: 的结构识别,发布后仍要扫描 md 文件是否残留单星标签
  • Markdown 转换不能只匹配小写 org——技能正文常写 Org、内联 #+description,Org 模板的 # 注释 在 Markdown 里还会变成标题;转换器统一处理这些形式,并在 md commit 前扫描输出指令、Org 标记和残留 .org 文件
  • 整份文档模板不能只改围栏标签——```org 或无语言围栏里的 #+title、星号标题与 Org 链接必须作为一个语义单元转换;转换器会生成带 --- 的 YAML、Markdown 标题与链接,同时保留原文件的纯 CRLF/LF 换行风格,避免无意义的全文件 diff
  • 输出格式残留不能只扫动词短语——「写 Org 文件时」转掉后,同段的「所有生成的 Org 文件」仍可能留下,令 md 分支继续要求生成 Org。转换器和提交前审计都要覆盖这种保存路径句式;否则一次成功 push 会把旧格式契约重新带回远端
  • 输出契约不只在 Markdown 文档里——evals/*.json 的 prompt / expected_output 会直接要求技能生成某种格式,运行时代码里的 stdin__*.org 还会选择默认解析器。mdize 与提交前审计必须覆盖这两类消费者,同时保留显式 Org 输入兼容测试
  • 互斥格式句不能只替换格式名——一句同时声明「必须使用源格式」和「禁止目标格式」时,逐词替换会让肯定项与禁止项重合。mdize 必须整句交换目标与排除项,残留审计也必须拒绝两端相同的句子
  • 格式名变了不等于语法规则也变了——org-mode 被替换为 markdown 后,单星号加粗、~code~ 与「不用反引号」仍会命令模型输出 Org。转换器必须把这组互斥规则整体改成 Markdown 规则,审计也要把旧规则视为阻断
  • 脚本会自动 bump patch version 在 plugin.json + marketplace.json——如果你想 bump minor / major,先手动改完再跑脚本,脚本只追加 patch
  • 如果 md 分支的远端比本地新(继刚另一台机器推过),脚本会 pull --rebase 失败时尝试一次 reset --hard origin/md 重新应用——这会丢弃本地未推的 md 分支 commit。脚本前会提示
  • 当前路径:skill 源固定在 ~/.agents/skills/,工作 repo 固定在 ~/code/ljg-skills/;不要从历史备份目录读取或推送

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Ljg Push AI skill do?

把 ~/.agents/skills/ljg-* 里所有更新过的 skills 同步到 github repo (ljg-skills),先推 master 分支(org-mode 输出风格),再切 md 分支(markdown 输出风格)做基础 markdown 化后推。Use when user says '/ljg-push', 'push skills', '推送 skills', '同步 skills', 'sync ljg', or whenever ljg-* skills get updated and need shipping. NOT FOR pushing non-ljg skills or arbitrary git repos.

Why use Ljg Push on TypingMind?

Because you install it once and use it with any model. Ljg Push is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Ljg Push in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/lijigang/ljg-skills/tree/master/skills/ljg-push. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Ljg Push?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Ljg Push?

As many as you like. As long as a model supports skills, you can use Ljg Push with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Ljg Push AI skill free?

Yes. It is published on GitHub by lijigang under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇