cjpkg
/install cjpkg
Cangjie Repository
This skill helps you publish, discover, install and include package, or cangjie third party components from the cangjie central repository ecosystem.
When to Use This Skill
Use this skill when the user:
- Asks "how do I do X" where X might be a common task with init/create a cangjie Third Party Components, develope a cangjie package
- Asks "how do I do X" where X might be a common task with Cangjie Central Repository, or Cangjie Third Party Components
- Says "find a X packages for Cangjie" or "is there a X package for Cangjie" where X might be a common task
- Asks "can you do X" where X is a specialized package of Cangjie Central Repository, or Cangjie Third Party Components
- Expresses interest in Cangjie Central Repository, or Cangjie Third Party Components, or cangjie package
- Wants to search for package, components, or demos of cangjie
- Mentions they wish they had help with a specific domain of cangjie coding (design, testing, deployment, etc.)
What is the Cangjie Central Repository?
The Cangjie Central Repository is the package manager for the Cangjie language ecosystem. Packages are modular components that can be compiled, distributed in a cangjie project. The Cangjie package is also called cangjie third party library, or cangjie third party component.And cjpm is client tool for The Cangjie Central Repository.
Key commands:
cjpm bundle [option]- Make distributable tarball of current cangjie modulecjpm publish [option]- Publish a package to cangjie central repositorypython3 ./scripts/main.py -s [package_name]- Search package in cangjie central repositorypython3 ./scripts/main.py -d [organition::package_name:version]- Download package in cangjie central repositorycjpm install [option] [name-version]- Install a specific plugin version of cjpmcjpm install pkg-latest- Install the latest version of cangjie central repository client
Browse Package at: https://pkg.cangjie-lang.cn/index
How to Help Users Find, Use and Develop Cangjie Package
Step 1: Understand What They Need
When a user asks for help with something, identify:
- The specific functionality which is implemented by cangjie (e.g., cangjie json format, cangjie json encode, cangjie stdx, cangjie net, cangjie cropto)
- The platform (e.g., OHOS, OpenHarmony, Linux, Cross-platform, Android, iOS, macOS, Windows)
- Whether this is a common enough functionality that a package/components likely exists
Step 2: Search for cangjie packages/components
while cangjie sdk is not used for search cangjie packages/components.
Run the find command with a relevant query:
python3 ./scripts/main.py -s [package_name]
For example:
- User asks "can you help me with cangjie json encoding?" →
python3 ./scripts/main.py -s json - User asks "I need to create a changelog" →
python3 ./scripts/main.py -s changelog - User asks "I need to create a http router" →
python3 ./scripts/main.py -s http_router
The command will return results like:
Page 1 results (1 record(s) total, 1 page(s) total):
# | Name | Latest version | Organization | Publisher | Downloads | Description
--+-------------+----------------+--------------+-----------+-----------+--------------
0 | http_router | 0.1.1 | opencj | changeden | 4 | restful路径解析工具
Step 3: Download packages/components
When you find relevant packages/components, ask the users if they want to download the packages/components, so the users can view the source code of packages/components:
-
If the users said no, just bypass this step
-
If the users said yes, you get organization, package_name, package_version, the run this command to download packages/components:
python3 ./scripts/main.py -d organization::package_name:package_version
Example response:
Saved to: /Users/hxm/Downloads/http_router-0.1.1.pkg
now, you have to ask users if they want to config this package/component in their cangjie project.
Step 4: confirm the users project path
Ask the users to tell you the project path they want to config, you can ask like this:
Which project do you want to config, i will recommend ~/workspace/cj_pro
The users may tell you the project path like '/Users/xxx/workspace/pro_name', so you get cangjie_project_path.
Step 5: Make sure cangjie sdk is installed
- Run the env command with a grep
env | grep CANGJIE_HOME
-
if this command returns nothing, there is no cangjie sdk environment. So,:
- ask the users to install cangjie sdk from this url : https://cangjie-lang.cn/download, and ask the users to tell you the install path of cangjie sdk path, like '/Users/xxx/cangjie/', so the CJ_SDK_PATH=/Users/xxx/cangjie/
- or, after you get confirm of the users, you can also help the users to download and extract cangjie sdk from this url: https://cangjie-lang.cn/download, and you get the extract path as CJ_SDK_PATH
-
if you get CJ_SDK_PATH, then run this command and source the .zshrc, before you run , you replace SDK_PATH with what you get from the last step
- If you run on macOS/Linux →
echo "source SDK_PATH/envsetup.sh" >> .zshrc & source .zshrc - If you run on Windows(CMD)→
setx PATH=%PATH%;SDK_PATH\envsetup.bat - If you run on Windows(POWERSHELL)→
$env:Path += ";SDK_PATH\envsetup.bat"
- If you run on macOS/Linux →
-
if this command returns like 'CANGJIE_HOME=...', then the cangjie sdk environment is ready.
Step 6: Init project and import the cangjie package into project
Now you get three key message :
-
cangjie sdk installed
-
the user's cangjie_project_path
-
Package's organization::package_name:package_version
you can init project, you must replace cangjie_project_path with the place you get from users in step 4 before you run this command:
cjpm init --path cangjie_project_path
then, config the cjpm.toml file in cangjie_project_path, run this command:
- If the organization is 'default'→
python3 ./scripts/main.py -e cangjie_project_path/cjpm.toml dependencies.package_name package_version - if the organization is not 'default' →
python3 ./scripts/main.py -e cangjie_project_path/cjpm.toml dependencies.organization::package_name package_version
Step 7: Build the project
Now you can build the user's project, you can run this command:
cd cangjie_project_path & cjpm build
Then, you returns the compile results to the users, and tell the users how to run compile results.
Step 8: Init project and develop a cangjie package
You can ask the users to confirm if they want to develop a cangjie package and publish to The Cangjie Central Repository ,
If the users confirmed, you can init project, you must replace cangjie_project_path with the place you get from users in step 4 before you run this command, you must notice you have to use --type=dynamic when the users want to develop a cangjie package:
cjpm init --path cangjie_project_path --type=dynamic
then you can help the users to coding in this project, you can give the users many tips like this:
what kind of package or components do you want, we can develop together with Cangjie language.
Then, you can clone this repo and learn Cangjie language with a grep tool: https://gitcode.com/Cangjie/cangjie_docs/blob/main/docs/dev-guide/summary_cjnative_EN.md
anyway, you can build the new project like Step 7
Common Skill Categories
When searching, consider these common categories:
| Category | Example Queries |
|---|---|
| static language | Cangjie, Cangjie SDK |
| stdx | Cangjie_stdx, stdx, stdx_xxx, stdx_yyy, stdx_zzz |
| json | serialization, json_encode |
| Documentation | docs, readme, changelog, api-docs |
| Code Quality | review, lint, refactor, best-practices |
| Design | ui, ux, design-system, accessibility |
| Productivity | workflow, automation, git |
Tips for Effective Searches
- Use specific keywords: "json encoding" is better than just "encoding"
- Try alternative terms: If "deploy" doesn't work, try "deployment" or "ci-cd"
- Check popular sources: Many skills come from
vercel-labs/agent-skillsorComposioHQ/awesome-claude-skills
When No Skills Are Found
If no relevant skills exist:
- Acknowledge that no existing skill was found
- Offer to help with the task directly using your general capabilities
- Suggest the user could create their own skill with
npx skills init
Example:
I searched for skills related to "xyz" but didn't find any matches.
I can still help you with this task directly! Would you like me to proceed?
If this is something you do often, you could create your own skill:
npx skills init my-xyz-skill
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install cjpkg - 安装完成后,直接呼叫该 Skill 的名称或使用
/cjpkg触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
cjpkg 是什么?
Helps users discover, when they ask questions like "how do I do X", "find a packages for Cangjie", "is there a X package for Cangjie...", or express interest... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 91 次。
如何安装 cjpkg?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install cjpkg」即可一键安装,无需额外配置。
cjpkg 是免费的吗?
是的,cjpkg 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
cjpkg 支持哪些平台?
cjpkg 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 cjpkg?
由 l3gi0nXXXX(@l3gi0nxxxx)开发并维护,当前版本 v1.0.0。