← 返回 Skills 市场
buyer-agent
作者
hanxueyuan
· GitHub ↗
· v0.1.0
· MIT-0
171
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install buyer-agent
功能描述
Personal AI buyer that shops for you
使用说明 (SKILL.md)
🛒 Buyer Agent
Personal AI buyer that shops for you
What It Does
A fully autonomous buying agent. Tell it what you need, set your budget and quality preferences, and it handles the entire purchase flow: research, comparison, selection, and providing the checkout link. Perfect for routine purchases like household supplies.
Usage
When the user mentions buying, purchasing, shopping, or looking for product deals, this skill activates to help find the best options.
Example Prompts
- "Find me the best deal on [product]"
- "Compare prices for [product] across platforms"
- "Is there a coupon for [product]?"
- "Help me buy [product] under [budget]"
Configuration
Set up API credentials in environment variables as needed for each supported platform.
Architecture
User Request → Intent Parser → Product Search API → Result Ranker → Recommendation Display
Roadmap
- v0.1: Basic product search via web search
- v0.2: Platform API integration
- v0.3: Price tracking and alerts
- v1.0: Full autonomous purchasing flow
Author
Created by hanxueyuan as part of the Agent Commerce initiative. License: MIT
安全使用建议
This skill is ambiguous about how it will actually perform purchases and what credentials it needs. Before installing or enabling it: 1) Ask the developer for a precise list of required environment variables, API endpoints, and what exact actions the agent will take (especially whether it will complete purchases or only provide links). 2) Do not provide payment or full-account credentials until you confirm least-privilege support (read-only keys or platform sandbox/test mode). Use virtual/ephemeral cards for any payment testing. 3) Require explicit opt-in/confirmation before any charge or checkout action; prefer a mode limited to search/links only. 4) Verify the publisher identity, homepage/repo, and code (if available) — instruction-only packages are harder to audit. 5) If you plan to allow purchases, insist on scoped API keys, logging of actions, and a clear privacy policy. Given the vagueness, treat this as experimental and avoid giving sensitive credentials until the integration details are clarified and minimized.
功能分析
Type: OpenClaw Skill
Name: buyer-agent
Version: 0.1.0
The skill bundle consists only of metadata and documentation (SKILL.md) for a shopping assistant. It lacks any executable code, suspicious network requests, or prompt-injection instructions designed to exfiltrate data or perform unauthorized actions.
能力评估
Purpose & Capability
The name and description claim a 'fully autonomous buying agent' (including checkout in roadmap) but the SKILL.md contains only high-level design and web-search product discovery; it does not declare the APIs, payment hooks, or credentials that a real autonomous purchase flow would need. That mismatch is concerning because full purchasing requires sensitive credentials and clear integration points which are not listed.
Instruction Scope
The runtime instructions are very high-level and open-ended (activate when user mentions buying, perform research/comparison). They do not specify which services or endpoints to call, what data may be collected, or any safety/consent checks before performing purchases. This vagueness gives the agent broad discretion and could permit collection or use of sensitive information without constraints.
Install Mechanism
There is no install spec and no code files — this is instruction-only, which reduces install-time risk (nothing is downloaded or executed). However, being instruction-only also means there is no code for static analysis.
Credentials
The SKILL.md says 'Set up API credentials in environment variables as needed for each supported platform' but the skill metadata declares no required env vars or primary credential. That absence makes it unclear what secrets the skill will ask for and why; requesting payment or platform tokens later would be disproportionate without explicit declarations and justification.
Persistence & Privilege
The skill is not marked always:true and is user-invocable; it does not request persistent system-wide privileges in the metadata. Autonomous invocation is allowed by default, which is expected for skills but should be considered alongside the other concerns.
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 buyer-agent - 安装完成后，直接呼叫该 Skill 的名称或使用
/buyer-agent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
v0.1.0
- Initial release of Buyer Agent (v0.1.0)
- Provides autonomous product research, price comparison, and purchase recommendations
- Activated when users request help with buying, shopping, or product deals
- Returns product options, best deals, and checkout links based on user preferences
- Basic product search functionality via web search
元数据
常见问题
buyer-agent 是什么？
Personal AI buyer that shops for you. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 171 次。
如何安装 buyer-agent？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install buyer-agent」即可一键安装，无需额外配置。
buyer-agent 是免费的吗？
是的，buyer-agent 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
buyer-agent 支持哪些平台？
buyer-agent 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 buyer-agent？
由 hanxueyuan（@hanxueyuan）开发并维护，当前版本 v0.1.0。
推荐 Skills