← 返回 Skills 市场
1380
总下载
1
收藏
3
当前安装
1
版本数
在 OpenClaw 中安装
/install vast-ai
功能描述
Provision and manage on-demand GPUs on VAST.ai, including search by GPU and price, renting containers, retrieving SSH, and checking account balance.
安全使用建议
What to consider before installing/use: 1) The skill legitimately targets VAST.ai and only needs your VAST API key to operate, which is proportional. 2) The package metadata omits the required VAST_API_KEY — the SKILL.md and code do require it; treat that omission as a red flag and ask the author to correct the metadata. 3) SKILL.md references an absolute developer path (/Users/sschepis/Development/...) and assumes a built dist/cli.js exists; the repo only contains TypeScript sources. You will need to run npm install and build (npx tsc) before the CLI will work — building/installation runs third-party code, so review compiled JS before executing and consider running in an isolated environment. 4) Before providing your VAST API key, verify the network endpoints in the code (baseUrl is console.vast.ai/api/v0) and ensure there are no hidden/external endpoints; this code appears to use only axios to the VAST API. 5) Because the skill can perform write actions (rent, destroy), only supply a key with least privilege (if VAST supports restricted keys) or use an account with limited funds. 6) Ask the publisher to: (a) declare VAST_API_KEY in the skill metadata, (b) remove absolute local paths from SKILL.md and use relative or packaged dist paths, and (c) include built artifacts or an install script so the agent/runtime behaviour is reproducible. If you cannot verify these fixes, run the code in an ephemeral VM/container and avoid giving your production VAST API key.
功能分析
Type: OpenClaw Skill
Name: vast-ai
Version: 1.0.0
The skill bundle is benign. The `SKILL.md` instructions are clear, task-oriented, and fully aligned with the stated purpose of managing VAST.ai GPU rentals, showing no signs of prompt injection. The TypeScript code (`src/*.ts`) implements a client for the legitimate VAST.ai API (`https://console.vast.ai/api/v0`), using standard libraries (`axios`, `limiter`, `minimist`) and handling API keys securely via environment variables. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or obfuscation. The absolute path in `SKILL.md` for execution is likely a developer's local path and not indicative of malicious intent.
能力评估
Purpose & Capability
The repository code and SKILL.md implement a VAST.ai client (search, rent, connect, balance) which matches the skill name. However the skill registry metadata claims no required environment variables while the runtime requires a VAST API key (VAST_API_KEY). That mismatch is unexpected and incoherent with the advertised metadata.
Instruction Scope
SKILL.md instructs the agent to run a CLI at an absolute developer path (node /Users/sschepis/Development/vast-ai/dist/cli.js) which will not exist in a runtime/agent environment. The instructions otherwise are scoped to VAST actions and ask the user for an API key before write operations, which is appropriate, but the absolute path and the assumption that a built dist exists are problematic.
Install Mechanism
There is no install spec (instruction-only), but full TypeScript sources, package.json and package-lock.json are included. Running this requires npm install + build (npx tsc) to create dist/*. No external download URLs or shorteners are used; dependencies are standard npm packages (axios, limiter, minimist). The lack of an install step in the skill metadata combined with source-only content is a usability/security concern (you must build/install to run).
Credentials
Runtime requires a single VAST API key (VAST_API_KEY / context.API_KEY) which is proportional to the stated purpose. However the registry metadata lists no required environment variables or primary credential, and SKILL.md/CLI both require VAST_API_KEY. This omission is a significant inconsistency and could lead to accidental exposure or misconfiguration.
Persistence & Privilege
The skill does not request permanent 'always' presence, does not modify other skills or system-wide settings, and does not declare elevated privileges. Autonomous invocation is enabled (default) which is normal and not a standalone concern.
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 vast-ai - 安装完成后，直接呼叫该 Skill 的名称或使用
/vast-ai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 release of the VAST.ai GPU Rental skill.
- Provision on-demand GPU infrastructure via VAST.ai.
- Search for machines by GPU type and price.
- Rent containers on selected offers.
- Retrieve SSH connection info for active containers.
- Check account credit balance and hourly burn rate.
- Implements step-by-step workflow, including pre-flight balance checks and user warnings for low credit.
元数据
常见问题
Vast Ai 是什么？
Provision and manage on-demand GPUs on VAST.ai, including search by GPU and price, renting containers, retrieving SSH, and checking account balance. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1380 次。
如何安装 Vast Ai？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install vast-ai」即可一键安装，无需额外配置。
Vast Ai 是免费的吗？
是的，Vast Ai 完全免费（开源免费），可自由下载、安装和使用。
Vast Ai 支持哪些平台？
Vast Ai 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 Vast Ai？
由 sschepis（@sschepis）开发并维护，当前版本 v1.0.0。
推荐 Skills