← 返回 Skills 市场
97
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install nexus-edge-deployer
功能描述
Deploy 1-bit quantized AI models on VPS for Agent-as-a-Service with 98% margins.
安全使用建议
Do not install or run this skill without clarification. Ask the author for an explicit list of required credentials (Hetzner API token, SSH key locations, model download endpoints), the exact binaries/tools it will call (Ollama, llama.cpp, monitoring agents), and any cloud-init scripts or templates to be used. If you must test it: use a throwaway/test Hetzner project with minimal-permission API tokens, review any cloud-init/user-data and automation scripts before execution, run in an isolated account or sandbox, and ensure model files come from trusted sources. The current metadata is inconsistent with the runtime actions described — that gap should be resolved before granting the skill access to credentials or letting it run autonomously.
功能分析
Type: OpenClaw Skill
Name: nexus-edge-deployer
Version: 2.1.0
The skill bundle describes a legitimate utility for deploying and managing 1-bit quantized AI models on Hetzner VPS infrastructure. The instructions in SKILL.md focus on MLOps tasks such as provisioning via cloud-init, resource monitoring, and unit economics reporting, with no evidence of malicious intent, data exfiltration, or harmful prompt injection.
能力评估
Purpose & Capability
The SKILL.md explicitly requires provisioning Hetzner VPSes via the cloud API, pre-loading models, and configuring Ollama/llama.cpp; yet the registry metadata lists no required environment variables, no required binaries, and no config paths. A legitimate deployer would normally require at minimum a Hetzner API token (or SSH keys), model artifacts or download endpoints, and the inference binaries (Ollama/llama.cpp) or install instructions.
Instruction Scope
Instructions direct the agent to provision servers, deploy personas, benchmark, and configure monitoring/auto-scaling. They are high-level and grant broad discretion (e.g., 'Provision VPS via Hetzner API with cloud-init') but provide no concrete safe defaults, no explicit credential handling, and no locations for cloud-init/model artifacts. That vagueness can lead an agent to seek credentials or run network actions not scoped in the metadata.
Install Mechanism
This is an instruction-only skill (no install spec), which is lower-risk in that it doesn't ship or execute code on install. However, the instructions implicitly require installing/using external tools (Ollama, llama.cpp, cloud-init, monitoring agents) but provide no guidance. The absence of an install spec is inconsistent with the operational requirements described.
Credentials
The skill declares no required environment variables or primary credential, but its workflow explicitly calls for Hetzner API provisioning and likely needs SSH keys, API tokens, or cloud credentials and access to model files. This mismatch is disproportionate and should be resolved before trusting the skill to act.
Persistence & Privilege
Flags show always:false and normal autonomous invocation settings. The skill does not request persistent presence or manipulate other skills' configs. No elevated platform privileges are declared.
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 nexus-edge-deployer - 安装完成后，直接呼叫该 Skill 的名称或使用
/nexus-edge-deployer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
v2.1.0
Summary: Version 2.1.0 introduces a detailed workflow, new model support, monitoring, and enhanced deployment features.
- Added support for deploying 1-bit quantized AI models (PrismML Bonsai, Microsoft BitNet) on minimal VPS infrastructure.
- New workflow section outlines client assessment, model selection, VPS provisioning, deployment, benchmarking, monitoring, and reporting.
- Enables calculation of Agent-as-a-Service (AaaS) unit economics, including cost, margin, and break-even analysis.
- Integrates Ollama or llama.cpp for multi-tenant inference serving and provides configuration guidelines.
- Automatic provisioning of Hetzner CX22 servers via Cloud API, with local inference for GDPR/HIPAA compliance.
- Introduces fleet monitoring: RAM, CPU, and tokens/sec per agent, plus scaling guidelines for up to 100+ agents.
- Offers best practices for benchmarking, agent density, security, and reporting cost savings.
元数据
常见问题
nexus-edge-deployer 是什么？
Deploy 1-bit quantized AI models on VPS for Agent-as-a-Service with 98% margins. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 97 次。
如何安装 nexus-edge-deployer？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install nexus-edge-deployer」即可一键安装，无需额外配置。
nexus-edge-deployer 是免费的吗？
是的，nexus-edge-deployer 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
nexus-edge-deployer 支持哪些平台？
nexus-edge-deployer 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（macos, linux, windows）。
谁开发了 nexus-edge-deployer？
由 Shuwanito（@shuwanito）开发并维护，当前版本 v2.1.0。
推荐 Skills