← 返回 Skills 市场

Kimi Search 极简版

Name: Kimi Search 极简版
Author: jhc888007

作者 jhc888007 · GitHub ↗ · v1.0.10 · MIT-0

cross-platform ✓ 安全检测通过

185

总下载

当前安装

版本数

在 OpenClaw 中安装

/install kimi-search-simple

功能描述

通过 Kimi API 的 builtin_function $web_search 联网检索。用于新闻、实时信息、股价、公司动态等时效性内容；支持中英检索。先判断用户要「原始搜索结果条目」还是「综合总结」（默认总结），再调用脚本或相应参数。

使用说明 (SKILL.md)

Kimi 联网搜索

用 Kimi 内置工具 $web_search 检索互联网最新信息。

前置准备

在调用脚本前，您需要在 Moonshot 开放平台申请 API Key。 https://platform.moonshot.cn/

请将 API Key 配置在系统的环境变量中。脚本会自动读取该变量（亦支持 KIMI_API_KEY）：

# Linux / macOS
export MOONSHOT_API_KEY="your_api_key_here"

# Windows (CMD)
set MOONSHOT_API_KEY="your_api_key_here"

意图识别（先做这个）

用户意图	表现	调用方式
总结（默认）	要结论、分析、简报、「帮我查一下并说明」	`search(query)` 或 `python3 scripts/kimi_search.py -q "..."`，不要加 `--raw`
原始结果	要来源列表、逐条链接、摘录、引用、材料汇总、不要一句话概括	`search(query, raw_output=True)` 或 `python3 scripts/kimi_search.py -q "..." --raw`

未明确说「要原文/链接/逐条」时，按总结处理。

前置条件

依赖：

pip install openai httpx

密钥（任选其一）：

环境变量：MOONSHOT_API_KEY 或 KIMI_API_KEY
文件：~/.config/moonshot/api_key 或 ~/.openclaw/credentials/moonshot-api-key

申请地址：https://platform.moonshot.cn/

用法

在 skill 根目录下执行（scripts/kimi_search.py 相对该目录）：

总结（默认）

python3 scripts/kimi_search.py -q "你的检索问题"

原始搜索结果条目

python3 scripts/kimi_search.py -q "你的检索问题" --raw

Python（当前工作目录为 scripts 时）

from kimi_search import search

print(search("快手 最新财报"))
print(search("某主题", raw_output=True))

行为说明

总结：系统提示为总结，模型在联网后给出综合回答与引用。
原始：工具往返后会把系统提示换为中文严格格式，要求逐条输出 Title / Date / URL / Summary，禁止一句话概括；temperature=0.3。
模型默认 kimi-k2-turbo-preview，max_tokens=8192，thinking 关闭。

故障排查

连接或 SSL：检查网络；可用 curl https://api.moonshot.cn/v1/models -H "Authorization: Bearer $MOONSHOT_API_KEY" 测通。

鉴权失败：在控制台核对密钥是否有效。

限流：间隔重试或缓存重复查询。

许可

MIT

安全使用建议

This skill appears to do what it says: run a Python script that calls the Moonshot/Kimi web-search API and returns either a summary or raw result list. Before installing or running it: 1) Be aware you must provide a Moonshot API key (MOONSHOT_API_KEY or KIMI_API_KEY) — the registry metadata did not list this, so set the env var or put the key in one of the two config paths the script will read. 2) The script requires Python packages openai and httpx (pip install openai httpx). 3) Review and confirm you trust the Moonshot platform (https://platform.moonshot.cn/) because your queries and API key are sent there. 4) If you are concerned about key exposure, inspect the script yourself to confirm it only sends the key to https://api.moonshot.cn/v1 (which it does) and does not transmit the key elsewhere. If anything about the source or the Moonshot endpoint is unfamiliar, treat the key as sensitive and avoid installing until you verify the provider.

功能分析

Type: OpenClaw Skill Name: kimi-search-simple Version: 1.0.10 The skill is a legitimate tool for performing web searches via the Moonshot (Kimi) API. The script `scripts/kimi_search.py` correctly handles API authentication by checking environment variables or local configuration files and communicates only with the official API endpoint (api.moonshot.cn). No evidence of data exfiltration, malicious execution, or prompt injection was found.

能力标签

requires-sensitive-credentials

能力评估

ℹ Purpose & Capability

The name/description (Kimi web search) match the provided code and instructions: the script calls Moonshot/Kimi API via an OpenAI-compatible client and uses a $web_search builtin. One inconsistency: the registry metadata lists no required env vars, but both SKILL.md and the script clearly require a Moonshot API key (MOONSHOT_API_KEY or KIMI_API_KEY) or config files. This is likely an omission in the metadata rather than a functional mismatch.

✓ Instruction Scope

SKILL.md and scripts instruct the agent to read only the Moonshot API key (env var or specific config files), install the openai and httpx packages, and call the Moonshot API. The instructions do not ask for unrelated system files, broad environment scraping, or sending data to unexpected endpoints.

✓ Install Mechanism

There is no automated install spec; it's instruction-only and includes a simple pip dependency list (openai, httpx). That is low risk and proportionate for a Python script that performs HTTP calls.

ℹ Credentials

The script legitimately needs one API credential (MOONSHOT_API_KEY or KIMI_API_KEY) and will also read two specific config file paths (~/.config/moonshot/api_key, ~/.openclaw/credentials/moonshot-api-key). This credential scope is proportional to a web-search integration. The only concern is the metadata doesn't declare these required env vars — verify that you are comfortable providing your Moonshot API key.

✓ Persistence & Privilege

The skill does not request always:true, does not modify other skills or system-wide settings, and is user-invocable only. It does not persist elevated privileges on its own.

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 kimi-search-simple
安装完成后，直接呼叫该 Skill 的名称或使用 /kimi-search-simple 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

v1.0.10

- 优化了 API Key 配置说明，增加 Moonshot 平台申请与环境变量设置的具体指引。 - 文档结构微调，方便用户快速定位配置步骤和启动脚本方法。 - 进一步明确前置准备与意图识别流程，提升易用性。

v1.0.9

- Skill renamed to kuni-web-search, now using Kimi API's $web_search for web queries. - Tavily API and related code removed. - Adds intent recognition to decide between summary (default) or raw source output. - New script: scripts/kimi_search.py; old Tavily script and README removed. - Updated documentation for usage, dependencies, and API key configuration. - Supports both summarized results and original search entries based on user request.

v1.0.8

Skill has migrated from Kimi API to Tavily API for web search. - Replaced kimi_search.py with tavily_search.py using Tavily API. - Updated environment variable requirement to TAVILY_API_KEY. - Expanded search options: added parameters for search depth, topic, time range, date filtering, max results, domain inclusion/exclusion, images, and output format. - Usage instructions and documentation updated to reflect Tavily’s interface and advanced features.

v1.0.7

- No changes detected in this version. - Functionality, usage, and documentation remain the same as the previous version.

v1.0.6

- Removed decorative emphasis (bold "注意：") from the API request notice for a cleaner description. - No functional or code changes; documentation wording simplified.

v1.0.5

- Updated description to note that this is a minimal, token-saving version. - Simplified the required environment variable listing. - Removed detailed feature descriptions and usage notes for conciseness. - Maintained installation, configuration, and licensing instructions.

v1.0.2

- Added KIMI_API_KEY as a supported environment variable alongside MOONSHOT_API_KEY for API access. - Updated documentation to include installation instructions for required Python dependencies (openai, httpx). - Clarified that the tool sends network requests to https://api.moonshot.cn. - Enhanced environment variable setup instructions with examples for better usability.

v1.0.1

- No changes detected in this version. - Documentation, usage, and functionality remain the same as the previous release.

v1.0.0

- Initial release of kimi-search skill. - Enables web search via Kimi API for up-to-date news, live data, stock prices, and company info. - Supports both Chinese and English search queries. - Returns AI-synthesized search results with references. - Usage instructions and environment variable setup included.

元数据

Slug kimi-search-simple

版本 1.0.10

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 9

常见问题