← 返回 Skills 市场
qingkongzhiqian

Groundapi Web Researcher

作者 qingkongzhiqian · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ✓ 安全检测通过
162
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install groundapi-web-researcher
功能描述
Deep web research assistant — search, scrape, check trending topics, get daily briefings, and synthesize information from multiple sources into a structured...
使用说明 (SKILL.md)

网络调研助手

当用户需要对某个话题做深入了解,或类似以下表达时自动触发:

  • "帮我调研一下 XXX"、"查一下 XXX 的情况"
  • "XXX 的最新进展是什么"、"帮我了解一下 XXX"
  • "对比一下 A 和 B"、"总结一下这个话题"
  • "帮我搜一下..."、"研究一下..."
  • "现在什么最火"、"今天有什么热点"

前置条件

本 Skill 依赖 GroundAPI MCP Server 提供的工具。确保已配置 GroundAPI MCP 连接:

{
  "mcpServers": {
    "groundapi": {
      "url": "https://mcp.groundapi.net/mcp",
      "headers": {
        "X-API-Key": "sk_gapi_xxxxx"
      }
    }
  }
}

执行流程

Step 1 — 拆解调研问题

将用户的调研需求拆解为 1-3 个搜索查询,覆盖不同角度。

示例:用户说"帮我调研一下固态电池的发展现状"

  • 查询 1:固态电池 技术进展 2026(技术角度)
  • 查询 2:固态电池 产业链 公司 量产(产业角度)
  • 查询 3:固态电池 市场规模 预测(市场角度)

Step 1.5 — 热度预判(可选)

调用 info_trending() 查看该话题是否在全网热搜中,了解当前舆论热度。 调用 info_bulletin() 获取每日新闻简报,看该话题是否在今日重点事件中。

如果话题正在热搜,优先使用 recency="oneDay" 获取最新信息。

Step 2 — 搜索

对每个查询调用 info_search(query="...", count=10, recency="oneMonth")

如果话题有时效性或正在热搜中,使用更短的 recency(oneWeekoneDay)。

Step 3 — 筛选与抓取

从搜索结果中挑选最相关、最权威的 3-5 个链接(优先选择:官方来源 > 权威媒体 > 行业报告 > 博客),调用 info_scrape(url="...") 获取全文。

如果某个 URL 抓取失败,跳过并使用搜索摘要代替。

Step 4 — 综合输出

## 🌐 调研报告:{话题}

### 概述
(2-3 句话的核心结论)

### 关键发现

**1. {角度一标题}**
- 要点 A(来源:XXX)
- 要点 B(来源:XXX)

**2. {角度二标题}**
- 要点 A(来源:XXX)
- 要点 B(来源:XXX)

**3. {角度三标题}**
- 要点 A(来源:XXX)
- 要点 B(来源:XXX)

### 总结与观点
(基于多源信息的综合判断,标注哪些是事实、哪些是推测)

### 信息来源
1. [标题](URL) — 简要说明
2. [标题](URL) — 简要说明
3. ...

对比调研模式

当用户要求对比两个或多个事物时(如"对比 React 和 Vue"),调整流程:

  1. 对每个对比对象分别搜索
  2. 抓取各自最相关的 2-3 篇内容
  3. 按统一维度做对比表格输出:
### 对比:A vs B

| 维度 | A | B |
|------|---|---|
| 维度1 | ... | ... |
| 维度2 | ... | ... |
| ... |

### 结论
(根据使用场景给出建议)

注意事项

  • 始终标注信息来源,不要把不同来源的信息混在一起而不注明
  • 区分"事实"(有明确数据支撑)和"观点"(来自分析/预测)
  • 如果搜索结果质量不佳或信息过少,如实告知用户,不要编造内容
  • 抓取网页时尊重内容:不要抓取明显需要付费阅读的内容
  • 输出语言跟随用户
安全使用建议
This skill appears internally consistent, but before installing: 1) Verify you trust https://groundapi.net and its MCP endpoint (review privacy, terms, and what data they retain). 2) Provide a scoped API key (GROUNDAPI_KEY) with minimal permissions if possible, and rotate/revoke it if you stop using the skill. 3) Be aware the skill performs web scraping via the GroundAPI service — avoid sending sensitive or private URLs you don't want processed by a third party. 4) Because the skill can be invoked by the agent, monitor agent activity and logs for unexpected requests to the MCP endpoint. If you need more assurance, ask the vendor for documented API behavior and a list of endpoints the MCP server will contact on your behalf.
功能分析
Type: OpenClaw Skill Name: groundapi-web-researcher Version: 1.1.0 The groundapi-web-researcher skill is a legitimate web research assistant designed to search, scrape, and synthesize information using the GroundAPI MCP server (groundapi.net). The instructions in SKILL.md outline a standard and transparent workflow for processing user queries, and there is no evidence of data exfiltration, malicious code execution, or harmful prompt injection. The requirement for a GROUNDAPI_KEY is consistent with the tool's stated functionality.
能力标签
requires-sensitive-credentials
能力评估
Purpose & Capability
Name/description (web research, searching, scraping, trending, briefings) match the declared requirement: a single GroundAPI API key and calls to GroundAPI MCP endpoints. No unrelated services, binaries, or unusual config paths are requested.
Instruction Scope
SKILL.md instructs the agent to call GroundAPI MCP methods (info_search, info_scrape, info_trending, info_bulletin) and to select/summarize sources. It does not instruct reading local files, other env vars, or sending data to third-party endpoints outside the GroundAPI MCP URL shown. It explicitly warns not to scrape paywalled content and to cite sources.
Install Mechanism
No install spec or code files are present; this is an instruction-only skill so nothing will be downloaded or written to disk by an installer.
Credentials
Only one credential is required (GROUNDAPI_KEY) and that aligns with calling a hosted GroundAPI MCP service. The SKILL.md example shows an API key header, which is consistent with the declared primaryEnv.
Persistence & Privilege
always:false and normal autonomous invocation settings. The skill does not request persistent system-wide changes or access to other skills' credentials.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install groundapi-web-researcher
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /groundapi-web-researcher 触发
  4. 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
v1.1.0
v1.1.0: Add trending topics and daily bulletin for research context, update MCP endpoint
v1.0.0
- Initial release of Deep Web Research Assistant powered by GroundAPI MCP tools. - Automatically breaks down user research requests into multi-angle search queries. - Searches, filters, and scrapes authoritative content from multiple web sources. - Synthesizes findings into structured reports or comparison tables with clearly cited sources. - Differentiates between facts and opinions, and supports both general research and comparison tasks. - Requires configuration of GROUNDAPI_KEY for full functionality.
元数据
Slug groundapi-web-researcher
版本 1.1.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

Groundapi Web Researcher 是什么?

Deep web research assistant — search, scrape, check trending topics, get daily briefings, and synthesize information from multiple sources into a structured... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 162 次。

如何安装 Groundapi Web Researcher?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install groundapi-web-researcher」即可一键安装,无需额外配置。

Groundapi Web Researcher 是免费的吗?

是的,Groundapi Web Researcher 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Groundapi Web Researcher 支持哪些平台?

Groundapi Web Researcher 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 Groundapi Web Researcher?

由 qingkongzhiqian(@qingkongzhiqian)开发并维护,当前版本 v1.1.0。

💬 留言讨论