← 返回 Skills 市场
152
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install our-autoskills
功能描述
AutoSkills - 智能AgentSkills自动检测与安装系统。自动检测项目类型（Web/Python/学术/金融等），推荐并安装对应的AgentSkills。触发词：autoskills、项目类型检测、智能安装。
安全使用建议
This SKILL.md is an instruction-only wrapper that tells an agent to run `npx our-autoskills` / `npm install -g our-autoskills`. That will download and run code from the npm registry — do not run it blindly. Before installing or authorizing this skill, verify the npm package and GitHub source: inspect the package contents, confirm the maintainer identity, and review what the package will read/modify (skills list, ClawHub/agent config). Ensure Node/npm are present in a safe environment (e.g., sandbox or VM), and require explicit listing of the exact package version, checksum, and any credentials that will be used. If you expect the skill to auto-activate other skills or access agent tooling, ask the maintainer to declare required binaries, config paths, and any tokens/permissions, or perform these activations manually after review.
功能分析
Type: OpenClaw Skill
Name: our-autoskills
Version: 1.0.0
The skill bundle functions as a wrapper to execute an external npm package via 'npx our-autoskills' (SKILL.md). This pattern encourages the AI agent to perform arbitrary code execution on the host system to 'detect' and 'install' software, which is a high-risk behavior. While the stated intent is a utility for project analysis, the lack of local implementation and the reliance on fetching/executing remote code from a third-party registry without verification is a significant security concern.
能力评估
Purpose & Capability
The SKILL.md claims automated detection and one‑click installation of AgentSkills (including auto-activating finance/marketing skills and integrating with 'Hunter Agent' / ClawHub). However the skill metadata declares no required binaries, no install spec, and no required env/config paths. The documentation instructs users/agents to run `npx our-autoskills` and `npm install -g our-autoskills` (which require Node/npm available and will download remote code) but Node/npm are not listed as required. The automatic activation of other skills (and checking ClawHub CLI) implies access to local agent state and possibly credentials, but none are declared.
Instruction Scope
SKILL.md instructs running npx/npm commands which will fetch and execute remote packages from the npm ecosystem; it also lists checks (Node version, ClawHub CLI, Skills directory, installed Skills). The file-pattern detection logic (package.json, requirements.txt, *.tex) is consistent with the stated purpose, but the instructions give the agent the authority to download and run external code and to inspect the repository filesystem — this is expected for a detector/installer but carries execution risk and should be made explicit.
Install Mechanism
There is no formal install spec in the metadata, but the SKILL.md relies on npx/npm to obtain and run the implementation. That means the runtime will perform network downloads and execute code from the npm registry (source and registry not pinned or shown). Instruction-only skills that tell agents to run npx present a high-risk install vector because arbitrary code is fetched and executed at runtime; the skill does not document or pin the exact package version or provide checksums.
Credentials
The skill declares no required environment variables or credentials, which is good surface-wise, but it's inconsistent with its stated behavior of auto-activating other Skills and integrating with external tooling (ClawHub, Hunter Agent, Jarvis-Invest, Sains CRM). Those actions commonly require tokens or CLI access. The lack of declared credentials or explicit prompts/flows for obtaining them is an omission that makes it unclear how activations will be authorized.
Persistence & Privilege
always:false and user-invocable:true (normal). The SKILL.md suggests modifying local state (installing skills, activating integrations) but does not declare an always:true or other elevated persistence. Autonomous invocation is allowed by default which increases blast radius if the installed package is malicious, but there is no direct request to permanently alter other skills' configuration in the metadata.
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 our-autoskills - 安装完成后，直接呼叫该 Skill 的名称或使用
/our-autoskills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
v1.0.0
AutoSkills 1.0.0 - 项目类型智能检测与Skills一键安装系统
- 支持自动检测项目类型（Web/Python/学术/金融/营销/媒体等）
- 根据项目类型，智能推荐并一键安装最优AgentSkills
- 提供快速命令行入口：检测、安装、Skills列表、系统检查
- 集成Hunter Agent，金融/营销项目可自动激活相关AgentSkills
- 支持多维度（文件/关键词）项目类型检测算法
- 内置27+款常用预配置AgentSkills，覆盖多种场景
元数据
常见问题
AutoSkills CLI 是什么？
AutoSkills - 智能AgentSkills自动检测与安装系统。自动检测项目类型（Web/Python/学术/金融等），推荐并安装对应的AgentSkills。触发词：autoskills、项目类型检测、智能安装。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 152 次。
如何安装 AutoSkills CLI？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install our-autoskills」即可一键安装，无需额外配置。
AutoSkills CLI 是免费的吗？
是的，AutoSkills CLI 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
AutoSkills CLI 支持哪些平台？
AutoSkills CLI 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 AutoSkills CLI？
由 Adam（@adamwgp）开发并维护，当前版本 v1.0.0。
推荐 Skills