← 返回 Skills 市场
chsengni

技能更新管理器

作者 Chsengni · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
123
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install skills-update-manager
功能描述
管理已安装技能的注册、版本跟踪与自动更新检查;当用户安装新技能、检查技能更新或配置更新策略时使用。
安全使用建议
What to consider before installing: - Review the included script (scripts/skill_registry.py) yourself before allowing the agent to run it. It performs file writes (skills_registry.json and a MEMORY.md at a path relative to the skill), network requests to GitHub/ClawHub, and may trigger git clone/downloads. - BACKUP your workspace/MEMORY.md before first use. The script claims to 'append' but uses 'w' and may overwrite that file; confirm the MEMORY_FILE path matches where you want configuration written. - Confirm where skills_registry.json will be stored (it's under the skill package parent directory) and whether that location is acceptable. - If you allow automatic update checks, be aware the agent will contact GitHub and fetch repositories (possible supply of arbitrary code). Only enable this for trusted skills and consider running updates in a sandboxed environment. - If you want safer behavior, modify the script to: (1) use a configurable MEMORY path (avoid hard-coded multi-parent traversal), (2) open MEMORY.md with append mode or perform a careful merge instead of blind 'w', and (3) avoid auto-generating GitHub URLs without user confirmation. Overall: functionally coherent for an update manager, but implementation bugs and file-write behavior make it suspicious until reviewed and adjusted to match the environment and your safety expectations.
能力评估
Purpose & Capability
Name/description match the code: the script extracts SKILL.md metadata, registers skills, and checks GitHub/ClawHub for releases. Network access to GitHub/ClawHub and parsing HTML/JSON is expected for this purpose. Minor mismatch: SKILL.md instructs a 'working directory MEMORY.md', but the script targets a specific MEMORY.md path computed relative to the script (SCRIPT_DIR.parent.parent.parent / 'workspace' / 'MEMORY.md'), which may not be the same file the user expects.
Instruction Scope
The SKILL.md tells the agent to read/write a MEMORY.md in the current working directory and to auto-extract metadata and run git clone / downloads. The actual script will read and may overwrite a MEMORY.md at a path relative to the installed script (not necessarily the agent's current working directory). The script's _init_memory_config claims to 'only append, not create' but opens the file with 'w' and writes existing_content + config_content (overwriting), so it can modify/replace a user's MEMORY.md unexpectedly. Automatic cloning/downloading of repos (as part of update/installation flows) is also part of the expected behavior but increases risk because it fetches and places remote code on disk.
Install Mechanism
There is no installer (instruction-only skill) which is lower risk; however, the bundle includes an executable Python script that will be run by the agent. No external package download or installer URL is specified. The presence of runnable code in the skill bundle means you should review the script before allowing execution.
Credentials
The skill requests no environment variables or credentials. It uses unauthenticated GitHub API endpoints and HTML scraping for ClawHub — appropriate for checking public releases. There are no unexpected credential requests. One behavior to note: when no update_url is provided the script auto-generates a GitHub URL (https://github.com/user/{name}), which could lead the agent to access arbitrary GitHub paths based on skill names.
Persistence & Privilege
always:false (good). But the script writes a skills_registry.json into the skill package area and will write to a MEMORY.md at a computed path. The script's documented behavior ('append only') contradicts its implementation (it may overwrite MEMORY.md). Writing/modifying a top-level agent MEMORY.md or workspace file is a persistent and potentially disruptive side-effect — this deserves caution and validation of the exact paths used in your environment.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install skills-update-manager
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /skills-update-manager 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.1
修复记忆路径错误
v1.0.0
Initial release of skills-update-manager — a centralized tool for skill registration, version tracking, and update management. - Register installed skills with automatic metadata extraction from SKILL.md. - Track version and source for each skill; supports registration from GitHub, ClawHub, or file uploads. - Check for updates automatically (configurable); fetch new versions from GitHub or ClawHub. - Edit update-checking settings via MEMORY.md or with a script. - List all registered skills and their update status. - Update local skill records after an upgrade.
元数据
Slug skills-update-manager
版本 1.0.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

技能更新管理器 是什么?

管理已安装技能的注册、版本跟踪与自动更新检查;当用户安装新技能、检查技能更新或配置更新策略时使用。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 123 次。

如何安装 技能更新管理器?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install skills-update-manager」即可一键安装,无需额外配置。

技能更新管理器 是免费的吗?

是的,技能更新管理器 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

技能更新管理器 支持哪些平台?

技能更新管理器 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 技能更新管理器?

由 Chsengni(@chsengni)开发并维护,当前版本 v1.0.1。

💬 留言讨论