← 返回 Skills 市场
minimax-tools
作者
Jiwei Wang
· GitHub ↗
· v1.0.0
· MIT-0
307
总下载
0
收藏
2
当前安装
1
版本数
在 OpenClaw 中安装
/install minimax-tools
功能描述
使用 MiniMax MCP 进行图像理解和网络搜索。触发条件：(1) 用户要求分析图片、理解图像 (2) 用户要求进行网络搜索、在线搜索 (3) 需要查询最新资讯、新闻、资料
安全使用建议
This skill's functionality is coherent, but the metadata omits two important runtime requirements: the 'uvx' binary (uv/uvx) and a MINIMAX_API_KEY. The install steps require executing a remote install script (astral.sh) and installing a third-party MCP package that will run with your API key. Before installing: (1) verify the source/trustworthiness of astral.sh and the minimax-coding-plan-mcp package (look for official project pages or signed releases), (2) avoid running curl | sh from unknown domains — prefer package managers or manual review, (3) give the MiniMax API key limited scope and do not reuse sensitive credentials, (4) insist the registry metadata be updated to declare required binaries and MINIMAX_API_KEY so you can make an informed consent decision. If you cannot verify the upstream project or prefer not to run remote installers, do not install this skill.
功能分析
Type: OpenClaw Skill
Name: minimax-tools
Version: 1.0.0
The skill bundle provides image understanding and web search capabilities by wrapping a MiniMax MCP server. It is classified as suspicious because the SKILL.md instructions include high-risk patterns such as pipe-to-shell installation (curl | sh for 'uv') and the installation of a third-party package (minimax-coding-plan-mcp) via uvx, which an AI agent might execute blindly. Additionally, the documentation includes a referral code (code=GjuAjhGKqQ) in the API registration link, which is atypical for standard utility skills. While the Python scripts (understand_image.py and web_search.py) use safe subprocess handling for JSON-RPC communication, the combination of unverified package installation and referral-based documentation warrants caution.
能力评估
Purpose & Capability
Name/description (image understanding + web search) matches the included scripts. However the registry/metadata claims no required env vars or binaries, while both SKILL.md and the scripts require an API key (MINIMAX_API_KEY or ~/.openclaw/config/minimax.json) and the 'uvx' binary (minimax-coding-plan-mcp) to operate. That mismatch between declared requirements and actual needs is incoherent.
Instruction Scope
SKILL.md instructs installing 'uv' via a curl | sh script and installing 'minimax-coding-plan-mcp' with uvx, then placing API keys in a local config file or env var. The runtime scripts read only the declared config/env and call an external MCP process; they don't access unrelated system files. The main scope concern is that instructions direct the user to run a remote installer and to hand an API key to an external binary, which expands the trust surface beyond the skill itself.
Install Mechanism
There is no formal install spec in the registry, but SKILL.md tells users to run 'curl -LsSf https://astral.sh/uv/install.sh | sh' to install uv and then 'uvx install minimax-coding-plan-mcp'. Downloading and executing a remote install script (astral.sh) is higher risk than using vetted package managers or published release artifacts. The MCP install is opaque (uvx-managed), so the skill effectively causes third-party code to be fetched and executed.
Credentials
The registry lists no required environment variables, but the scripts and SKILL.md clearly require MINIMAX_API_KEY (or a config file with an api_key). Requesting an API key for the service itself is reasonable, but the metadata omission is a red flag. Passing the API key to the external 'uvx' process (via env) means the external MCP will have access to the credential; consider whether that key has limited scope and whether you trust the upstream service.
Persistence & Privilege
The skill is not marked always:true and does not request unusual platform privileges. It writes/reads only within ~/.openclaw paths for its own config and workspace. The agent-autonomy default remains allowed (disable-model-invocation is false) but that is normal and not by itself a problem.
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 minimax-tools - 安装完成后，直接呼叫该 Skill 的名称或使用
/minimax-tools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
Supports MiniMax online search and image understanding (MiniMax API key configuration required). 支持MiniMax联网搜索、图片理解（需要配置MiniMax API key）
元数据
常见问题
minimax-tools 是什么？
使用 MiniMax MCP 进行图像理解和网络搜索。触发条件：(1) 用户要求分析图片、理解图像 (2) 用户要求进行网络搜索、在线搜索 (3) 需要查询最新资讯、新闻、资料. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 307 次。
如何安装 minimax-tools？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install minimax-tools」即可一键安装，无需额外配置。
minimax-tools 是免费的吗？
是的，minimax-tools 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
minimax-tools 支持哪些平台？
minimax-tools 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 minimax-tools？
由 Jiwei Wang（@oujakivi）开发并维护，当前版本 v1.0.0。
推荐 Skills