← 返回 Skills 市场
SuperClaw
作者
gvuckovich
· GitHub ↗
· v1.0.0
· MIT-0
165
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install superclaw-skills
功能描述
14 production-tested agent workflow skills for disciplined, high-quality AI coding and task execution. Covers TDD, debugging, code review, planning, brainsto...
安全使用建议
This bundle appears to be what it says: workflow skills plus a browser-based 'visual companion' implemented with a local Node HTTP/WebSocket server. Before installing or running it, consider: 1) Runtime requirements: the package includes Node.js (server.cjs) and Bash scripts — ensure your environment provides node and a POSIX shell (these are not declared in the metadata). 2) Filesystem & git access: the SKILL.md explicitly instructs reading project files, writing specs to docs/, and committing; only run this in repos and directories you trust. 3) Local server exposure: the visual companion can be bound to 0.0.0.0 and served for remote/containerized setups — prefer binding to localhost unless you intentionally want external access, and inspect start-server.sh/server.cjs before running. 4) Review server.cjs and helper.js for any network endpoints or unexpected behavior (the provided code uses local WebSocket/HTTP only). 5) If you will run the visual companion, prefer using a project-specific .superpowers directory (and add it to .gitignore) to avoid leaving ephemeral files in your repo root. If you want higher assurance, ask the author for declared runtime requirements (node version) and a signed release or run the code in an isolated environment/VM first.
能力评估
Purpose & Capability
The skill bundle claims to provide 14 agent workflow skills and the repository includes SKILL.md files and supporting scripts that implement those behaviors (e.g., a local visual companion with a Node HTTP/WebSocket server, shell scripts to start/stop it, TypeScript examples, helper JS). That is coherent with the description. Minor inconsistency: metadata lists no required binaries, but several delivered scripts assume Node.js and a POSIX shell (bash). The absence of declared runtime requirements (node, bash) is an omission but not evidence of malicious intent.
Instruction Scope
The SKILL.md instructions explicitly tell the agent to examine project files, write spec files to docs/, commit to git, and (for visual work) run a local server and read/write files under a session directory (SCREEN_DIR). Those actions are within the expected scope for brainstorming/planning code-workflow skills, but they do grant filesystem and git access and instruct running a local webserver and reading a generated '.events' file. This is expected for the visual companion feature but worth noting because it involves local I/O and starting processes.
Install Mechanism
There is no platform install spec; this is instruction-plus-source. That is low risk in terms of automatic downloads. The package ships Node and shell scripts (server.cjs, start-server.sh, etc.) which will be executed by the agent if you follow the SKILL.md guidance — the files are present and readable but not automatically installed from an external URL. No suspicious downloads or third-party registries are in the manifest.
Credentials
The skill declares no required credentials or env vars, and I see no code that requires cloud API keys. Some scripts respect several environment variables (BRAINSTORM_DIR, BRAINSTORM_HOST, BRAINSTORM_URL_HOST, BRAINSTORM_OWNER_PID, BRAINSTORM_PORT) but these are operational knobs for the local server. There are no requests for unrelated secrets. Still, the skill's runtime expects access to the local filesystem and git credentials (to commit specs), which you should only grant in trusted contexts.
Persistence & Privilege
always:false and the skill does not demand permanent global presence. The included scripts create per-session directories (default /tmp or project/.superpowers/brainstorm/) and write files there (HTML screens, .events, .server-info). That behavior is consistent with a local visual companion and does not modify other skills or system-wide agent settings.
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 superclaw-skills - 安装完成后，直接呼叫该 Skill 的名称或使用
/superclaw-skills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
Initial public release. 14 production-tested agent workflow skills.
元数据
常见问题
SuperClaw 是什么？
14 production-tested agent workflow skills for disciplined, high-quality AI coding and task execution. Covers TDD, debugging, code review, planning, brainsto... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 165 次。
如何安装 SuperClaw？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install superclaw-skills」即可一键安装，无需额外配置。
SuperClaw 是免费的吗？
是的，SuperClaw 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
SuperClaw 支持哪些平台？
SuperClaw 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 SuperClaw？
由 gvuckovich（@gvuckovich）开发并维护，当前版本 v1.0.0。
推荐 Skills