← 返回 Skills 市场
Solo CLI Guide
作者
Samarth Shukla
· GitHub ↗
· v1.0.3
· MIT-0
113
总下载
1
收藏
0
当前安装
4
版本数
在 OpenClaw 中安装
/install solo-cli-guide
功能描述
Interactive step-by-step tutor for Solo CLI — guides a human through environment setup, robot arm calibration, teleoperation, dataset recording, and policy t...
安全使用建议
This appears to be a coherent, human-in-the-loop tutor for Solo CLI. Before installing or running commands the tutorials show: (1) Review any remote installer scripts (e.g., astral.sh) before piping them to sh; (2) verify the GitHub repo (https://github.com/GetSoloTech/solo-cli) and PyPI package match the project you expect; (3) never paste tokens/credentials into chat — only enter them in your own terminal when the tooling asks; (4) pushing to HuggingFace or enabling W&B will require tokens you supply locally — the skill does not read or store them. If you need the agent to run commands autonomously, note that this skill explicitly does not and points to a different executor skill (solo_impl) which is not included here.
功能分析
Type: OpenClaw Skill
Name: solo-cli-guide
Version: 1.0.3
The skill bundle is an interactive tutor for the Solo CLI robotics platform, designed to guide users through environment setup, hardware calibration, and model training. While it includes potentially risky commands like 'curl | sh' for installing the 'uv' package manager (domains/environment.json), it explicitly instructs the AI agent to act only as a guide, forbids autonomous execution, and advises users to review scripts before running them. The commands are sourced from static JSON files with strict validation rules, and there is no evidence of data exfiltration, obfuscation, or malicious intent.
能力评估
Purpose & Capability
The name/description match the included domain actions, tutorials, and prompts. All required actions (installing uv, cloning solo-cli, running solo robo commands, pushing to HuggingFace/W&B) are coherent with a CLI tutor for Solo CLI; no unrelated credentials, binaries, or config paths are demanded.
Instruction Scope
SKILL.md and the tutor prompt confine the agent to bundled domain/tutorial files and to presenting commands verbatim from those files. The skill explicitly enforces human-in-the-loop validation and a hard boundary for out-of-scope questions. It asks the user to run environment checks in their own shell (e.g., VIRTUAL_ENV, groups $USER) but does not instruct the agent to read user files or secrets.
Install Mechanism
The skill is instruction-only (no install spec), which minimizes disk/write risk. However the tutorial instructs users to run networked installs: piping astral.sh install script to sh, cloning GitHub repos, and using uv pip install from PyPI. These are expected for this purpose but carry the usual risks of running remote install scripts and installing packages from the network — the skill itself warns users to review remote scripts before piping to shell.
Credentials
The skill declares no required environment variables or credentials. Optional credentials mentioned (HuggingFace token, Weights & Biases key) are only relevant to optional push/log flows and are justified by the training/publishing actions. The SKILL.md does ask users to check environment state themselves but does not request the agent access those values.
Persistence & Privilege
No always:true flag, no install spec, no code written to disk by the skill. The skill does not attempt to modify other skills or system-wide agent settings. Autonomous execution is disabled by policy in the prompt (this skill is a guide), so it has limited privilege.
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 solo-cli-guide - 安装完成后，直接呼叫该 Skill 的名称或使用
/solo-cli-guide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
v1.0.3
Clarify security section: agent reads only bundled files; validation steps ask the user to inspect their own env, not the agent
v1.0.2
Fix security section: clarify skill-level vs workflow-level network activity, declare optional HuggingFace and W&B credentials
v1.0.1
Rewrote SKILL.md to match ClawHub formatting standards — same functionality, cleaner structure
v1.0.0
- Initial release of Solo CLI Guide: an interactive, step-by-step tutor for environment setup, robot arm calibration, teleoperation, dataset recording, and policy training.
- Guides users through CLI operations using only commands sourced verbatim from included JSON domain files—no generated or inferred commands.
- Enforces strict validation after each step and addresses known issues before proceeding.
- Supports OS-specific instructions and references official documentation and community resources for deeper help.
- Does not perform actions autonomously; operates purely as a human-in-the-loop guide.
元数据
常见问题
Solo CLI Guide 是什么？
Interactive step-by-step tutor for Solo CLI — guides a human through environment setup, robot arm calibration, teleoperation, dataset recording, and policy t... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 113 次。
如何安装 Solo CLI Guide？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install solo-cli-guide」即可一键安装，无需额外配置。
Solo CLI Guide 是免费的吗？
是的，Solo CLI Guide 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Solo CLI Guide 支持哪些平台？
Solo CLI Guide 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 Solo CLI Guide？
由 Samarth Shukla（@samarthshukla6）开发并维护，当前版本 v1.0.3。
推荐 Skills