← 返回 Skills 市场
100
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install china-research
功能描述
国内社交媒体调研工具。触发条件:(1) 用户想了解某个领域的真实需求 (2) 用户想调研国内市场机会 (3) 用户想做产品需求验证 (4) 用户问"最近国内XX有什么动态" (5) 用户提及"国内调研"、"市场调研"、"用户反馈
使用说明 (SKILL.md)
china-research
国内社交媒体调研技能,给定话题后自动搜索国内主流平台的真实用户讨论,生成带引用的调研报告。
技术方案
通过 mcporter 调用 glm-web-search 或 tavily-search,使用 site: 指令间接搜索各平台内容,然后用 AI 整合生成报告。
执行流程
步骤 1: 确认搜索话题
向用户确认要调研的话题,并了解:
- 调研目的(产品验证/市场分析/竞品研究)
- 关注的侧重点(用户痛点/市场机会/现有方案)
步骤 2: 执行多轮搜索
对每个话题,分多轮搜索不同平台组合:
第1轮 - 通用搜索:
mcporter call glm-search.webSearchPrime search_query="\x3C话题> site:xiaohongshu.com OR site:zhihu.com OR site:weibo.com"
第2轮 - 痛点搜索:
mcporter call glm-search.webSearchPrime search_query="\x3C话题> 痛点 site:v2ex.com OR site:jike.city OR site:sspai.com"
第3轮 - 商业搜索:
mcporter call glm-search.webSearchPrime search_query="\x3C话题> site:36kr.com OR site:huxiu.com"
第4轮 - 技术搜索(如适用):
mcporter call glm-search.webSearchPrime search_query="\x3C话题> 推荐 site:zhihu.com OR site:v2ex.com"
每次搜索取 top 5 结果,总共最多 20 条来源。
步骤 3: 整合分析
根据搜索结果,按以下结构整理:
- 提取核心发现(3-5个要点)
- 归纳用户真实痛点
- 总结现有解决方案
- 分析市场空白与机会
步骤 4: 生成报告
按标准格式输出调研报告。
搜索策略模板
平台组合搜索
| 搜索目的 | 平台组合 | 搜索模板 |
|---|---|---|
| 大众反馈 | 小红书+知乎+微博 | \x3C话题> site:xiaohongshu.com OR site:zhihu.com OR site:weibo.com |
| 深度痛点 | V2EX+即刻+少数派 | \x3C话题> 痛点/抱怨/问题 site:v2ex.com OR site:jike.city OR site:sspai.com |
| 商业分析 | 36氪+虎嗅 | \x3C话题> site:36kr.com OR site:huxiu.com |
| 技术讨论 | 知乎+V2EX | \x3C话题> site:zhihu.com OR site:v2ex.com |
关键词变体
对同一话题,尝试不同关键词:
\x3C话题> 痛点\x3C话题> 体验\x3C话题> 推荐\x3C话题> 吐槽\x3C话题> 评测
输出格式
## 📋 [话题] 调研报告
**时间:YYYY-MM-DD | 数据来源:N条**
### 一、核心发现(3-5个要点)
- 发现1
- 发现2
- 发现3
### 二、用户真实痛点
(从知乎/V2EX/即刻提取的抱怨和需求)
- 痛点1:描述 + 来源
- 痛点2:描述 + 来源
### 三、现有解决方案
(市场上已有的产品/方案)
- 方案1:描述 + 来源
- 方案2:描述 + 来源
### 四、市场空白与机会
(有需求但解决方案不完善的领域)
- 机会1
- 机会2
### 五、参考来源
1. [标题](url) - 来源平台 - 摘要
2. [标题](url) - 来源平台 - 摘要
...
注意事项
- 来源必须可追溯:每个观点都要有来源链接
- 区分事实与观点:用户评论是主观观点,媒体报道是客观信息
- 时效性:优先引用近期内容(1年内)
- 平台特性:不同平台用户画像不同,结论要注明来源平台
参考文件
安全使用建议
This skill appears to do what it says (search Chinese social platforms and produce a sourced report), but before installing or enabling it confirm the following: (1) the runtime must provide 'mcporter' and the named connectors (glm-search.webSearchPrime, tavily-search) — ask the publisher which binaries/connectors are required or declare them in the skill metadata; (2) identify exactly which external services will receive your search queries (the connectors' endpoints) and review their privacy/retention policies — your search terms (possibly sensitive) will be sent to those services; (3) verify whether any API keys or credentials are needed for the connectors and where those secrets will be stored; (4) ensure the intended scraping/search behavior complies with target platforms' terms of service; and (5) if you require stronger privacy, consider running searches through a trusted, auditable search provider or doing manual sampling instead. If the publisher can confirm the mcporter/connectors are internal and that no external third-party endpoints or undisclosed credentials are involved, the remaining issues are lower risk.
功能分析
Type: OpenClaw Skill
Name: china-research
Version: 1.0.0
The china-research skill bundle is a legitimate tool designed for market research on Chinese social media platforms. It uses the mcporter tool to perform targeted web searches via glm-search.webSearchPrime on specific domains like zhihu.com, xiaohongshu.com, and v2ex.com. The logic in SKILL.md and references/platforms.md is transparent, lacks any malicious execution or data exfiltration patterns, and strictly follows its stated purpose of gathering and summarizing public user feedback.
能力评估
Purpose & Capability
The skill claims to perform web searches across Chinese platforms and its instructions use the 'mcporter' tool to call connectors (glm-search.webSearchPrime / tavily-search). However the skill metadata declares no required binaries or tools. Either the runtime must already provide mcporter and those connectors, or the skill is missing a declared dependency — this is an incoherence that affects whether the skill can perform its stated purpose.
Instruction Scope
Instructions are narrowly scoped to searching public platforms (site: queries), extracting top results, and producing a sourced report — which fits the stated purpose. However the SKILL.md explicitly routes all searches through third-party connectors (glm-web-search, tavily-search) invoked via 'mcporter', which means user queries and topics will be forwarded to whatever endpoints back those connectors. The skill does not document or disclose those endpoints, credentials, or privacy/retention behavior.
Install Mechanism
This is an instruction-only skill with no install spec and no code files — nothing will be written to disk by the skill package itself. That is low-risk, provided the runtime environment supplies the tools the instructions assume.
Credentials
The skill declares no environment variables or credentials, but its instructions rely on external search connectors that often require API keys or configuration. The metadata does not declare those credentials (primaryEnv or requires.env). This omission could hide required secrets or make behavior dependent on platform-specific connectors; users should verify what credentials the runtime's mcporter connectors need and where queries will be sent.
Persistence & Privilege
The skill does not request persistent presence (always: false) and has default autonomy settings. It does not attempt to modify other skills or system settings in the provided instructions.
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 china-research - 安装完成后,直接呼叫该 Skill 的名称或使用
/china-research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
首次发布:国内社交媒体调研技能,覆盖10个主流平台
元数据
常见问题
China Research 是什么?
国内社交媒体调研工具。触发条件:(1) 用户想了解某个领域的真实需求 (2) 用户想调研国内市场机会 (3) 用户想做产品需求验证 (4) 用户问"最近国内XX有什么动态" (5) 用户提及"国内调研"、"市场调研"、"用户反馈. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 100 次。
如何安装 China Research?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install china-research」即可一键安装,无需额外配置。
China Research 是免费的吗?
是的,China Research 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
China Research 支持哪些平台?
China Research 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 China Research?
由 云峰(@wuyunfeng8)开发并维护,当前版本 v1.0.0。
推荐 Skills