← 返回 Skills 市场
235
总下载
0
收藏
10
当前安装
4
版本数
在 OpenClaw 中安装
/install skill-discovery
功能描述
基于用户意图发现并推荐 skill。分析用户输入，搜索匹配 skill，验证质量后推荐最佳匹配，由用户确认后安装。
安全使用建议
This skill appears internally consistent and implements a recommendation-first workflow (dry-run) by default. Before installing or enabling it as an automatic hook, consider: 1) it will run 'npx skills' commands and may call the public ClawHub API (clawhub.ai) — ensure outbound network access to that host is acceptable; 2) it writes sanitized logs to $OPENCLAW_DIR/logs/skill-discovery-v3.json and stores uninstall backups in a .trash directory under the configured path (defaults to your home .openclaw) — verify the log/backup locations and retention meet your privacy/storage needs; 3) automatic installation only occurs if explicitly invoked with dryRun: false or explicit CLI flags — review the install command before consenting to install any discovered skill; 4) the code uses child_process.exec but implements shell escaping for user inputs; if you have strict security requirements, inspect the sanitize() implementation and the trustedOwners list to confirm they match your policy. Overall, nothing in the package requests unexplained credentials or hidden endpoints.
能力评估
Purpose & Capability
Name/description describe searching and recommending skills; the code implements searches via the official 'npx skills' CLI and an optional ClawHub API, sorts by installs/trusted owners, and exposes install/remove wrappers. Required binaries (npx, optional clawhub) and Node.js runtime are appropriate for this purpose.
Instruction Scope
Runtime instructions and code limit actions to: analyzing user input, calling 'npx skills' (find/list/add/remove), optionally calling the public ClawHub search API, building sanitized logs, and backing up/uninstalling installed skills. The skill documents and implements dry-run (recommendation) mode by default and does not perform automatic installs unless explicitly invoked with dryRun: false/explicit options.
Install Mechanism
No external binary downloads or obscure install URLs are used. The package is bundled (entrypoint index.js) and relies on existing CLIs (npx/clawhub) and Node.js. Source files are included; no extract-from-unknown-URL install steps are present.
Credentials
The skill requests no secrets and only uses optional environment variables (OPENCLAW_DIR, TRASH_DIR) to locate logs/backups; these are documented in SKILL.md. It does make outbound HTTPS requests to clawhub.ai (public search API) which is proportional to adding an alternate registry source.
Persistence & Privilege
The skill does not set always:true. It exposes an OpenClaw hook (onUserInput) which, when used, runs in dry-run mode by default. It writes logs and keeps uninstall backups under the user's OpenClaw directory (configurable). These behaviors are documented and proportional to its purpose.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install skill-discovery - 安装完成后，直接呼叫该 Skill 的名称或使用
/skill-discovery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
版本历史
v2.2.1
补充 install spec (type:bundled) 和 sanitize 脱敏策略文档，消除 OpenClaw ℹ 提示
v2.2.0
默认推荐模式，解决 OpenClaw Suspicious 标记：autoDiscover/onUserInput 默认 dryRun:true，skillsAdd 不再默认 -g -y
v2.1.1
修复安全扫描标记：排除 package-lock.json，声明依赖 CLI 工具
v2.1.0
添加 ClawHub 作为 skill 搜索源，与 skills.sh 并行搜索、合并去重
元数据
常见问题
Skill Discovery 是什么？
基于用户意图发现并推荐 skill。分析用户输入，搜索匹配 skill，验证质量后推荐最佳匹配，由用户确认后安装。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 235 次。
如何安装 Skill Discovery？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install skill-discovery」即可一键安装，无需额外配置。
Skill Discovery 是免费的吗？
是的，Skill Discovery 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Skill Discovery 支持哪些平台？
Skill Discovery 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 Skill Discovery？
由 caocong1（@caocong1）开发并维护，当前版本 v2.2.1。
推荐 Skills