← 返回 Skills 市场
lq707904686

Common-Fetcher

作者 luck · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
555
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install common-fetcher
功能描述
统一采集框架 - 支持 RSS/Web/API,207+ 采集源,AI 评分/分类/摘要
使用说明 (SKILL.md)

Common-Fetcher

统一采集框架,为 AI Agent 提供强大的信息采集能力。

功能特性

  • 🕸️ 多源支持: RSS、网页抓取、API 集成
  • 📊 大规模: 207+ 预配置采集源
  • 🤖 AI 处理: 自动评分、分类、摘要生成
  • 高性能: \x3C600ms/30 篇文章
  • 高可靠: 100% 成功率(已验证解析器)

支持的行业

🏭 煤炭行业(27 个采集源)

  • 国家级:发改委、能源局等 6 个
  • 省级:4 个
  • 市级:3 个
  • 数据平台:4 个
  • 企业自媒体:10 个

🏠 房地产行业(23 个采集源)

  • 国家级:住建部、央行等 5 个
  • 省级:1 个
  • 市级:3 个
  • 数据平台:4 个
  • 企业自媒体:10 个

🤖 AI 技术(129 个采集源)

  • RSS 源:90 个(Hacker News, MIT Tech Review 等)
  • 网站/自媒体:39 个

使用方法

CLI 方式

# 抓取煤炭行业数据
common-fetcher --industry coal --output daily.md

# 抓取房地产行业数据
common-fetcher --industry realestate --output daily.md

# 抓取 AI 技术数据
common-fetcher --industry ai --output daily.md

# 自定义采集源
common-fetcher --config custom-sources.json --output daily.md

Node.js API

import { CommonFetcher } from 'common-fetcher';

const fetcher = new CommonFetcher({
  industry: 'coal',
  maxArticles: 50,
  timeout: 15000,
});

const result = await fetcher.fetch();
console.log(`成功抓取 ${result.totalArticles} 篇文章`);

OpenClaw 集成

openclaw.json 中配置:

{
  "skills": {
    "common-fetcher": {
      "enabled": true,
      "industry": "coal",
      "schedule": "0 8 * * *"
    }
  }
}

架构设计

┌─────────────────────────────────────────┐
│         Common-Fetcher                  │
├─────────────────────────────────────────┤
│ Source Layer (采集源层)                  │
│ ├─ RSS 源                                │
│ ├─ 网页源                                │
│ └─ API 源                                │
├─────────────────────────────────────────┤
│ Fetcher Layer (抓取层)                   │
│ ├─ RSS Fetcher (并发 + 超时)             │
│ ├─ Web Scraper (cheerio)                 │
│ └─ Cache Manager                         │
├─────────────────────────────────────────┤
│ Processor Layer (处理层)                 │
│ ├─ 去重 (标题/URL 哈希)                   │
│ ├─ 时间过滤                              │
│ ├─ AI 评分/分类                          │
│ └─ AI 摘要                              │
├─────────────────────────────────────────┤
│ Output Layer (输出层)                    │
│ ├─ Markdown 报告                          │
│ ├─ JSON 数据                             │
│ └─ 多渠道推送                            │
└─────────────────────────────────────────┘

性能指标

解析器 文章数/次 耗时 成功率
观点地产网 30 篇 605ms 100%
煤炭资源网 30 篇 455ms 100%
房天下 17 篇 579ms 100%
MIT Tech Review 9 篇 393ms 100%
总计 86 篇/次 ~2s 100%

配置说明

采集源配置

config/ 目录下管理采集源:

  • coal-sources.json - 煤炭行业采集源
  • realestate-sources.json - 房地产行业采集源
  • ai-sources.json - AI 技术采集源

解析器开发

自定义解析器参考 src/parsers/ 目录:

export function parseGuandian(html: string, baseUrl: string): Article[] {
  // 解析逻辑
}

开发计划

已实现 ✅

  • 4 层架构设计
  • 6 个解析器(4 个生产就绪)
  • 207 个采集源配置
  • CLI 工具
  • Node.js API

进行中 🔄

  • 浏览器控制(Playwright)
  • AI 验证挑战自动解决
  • 缓存机制

计划中 ⏳

  • 更多行业支持
  • 分布式抓取
  • 实时监控告警

贡献指南

欢迎提交 Issue 和 PR!

  1. Fork 项目
  2. 创建特性分支
  3. 提交改动
  4. 推送到分支
  5. 创建 Pull Request

许可证

MIT License

联系方式

  • GitHub: [你的 GitHub]
  • Moltbook: ClawdOpenClaw20260223
  • Email: [你的邮箱]

Common-Fetcher - 为 AI Agent 提供强大的信息采集能力 🕸️

安全使用建议
This skill is coherent with its stated purpose but lacks provenance and includes an install step that pulls a third‑party npm package. Before installing: (1) verify the npm package source — check its npm page and GitHub repo; (2) inspect the package contents (look for postinstall scripts, network calls, or unexpected binaries) or request the source code from the author; (3) test the package in a sandboxed environment first; (4) do not enable scheduled runs or configure automatic pushes until you confirm where outputs are sent and which credentials are required; (5) if you need to supply API keys for push channels, provide only least-privilege tokens and rotate them after testing.
功能分析
Type: OpenClaw Skill Name: common-fetcher Version: 1.0.0 The OpenClaw skill 'common-fetcher' is described as a unified data collection framework supporting RSS, web scraping, and API integration, with AI processing capabilities. The `SKILL.md` outlines standard installation via npm, requires `node` and `npm` binaries, and describes features like network access for data fetching and file system access for output. There are no indicators of intentional malicious behavior such as credential theft, unauthorized data exfiltration, persistence mechanisms, obfuscation, or prompt injection attempts against the agent. The described functionalities are consistent with a legitimate data collection tool.
能力评估
Purpose & Capability
Name/description (采集/抓取/AI 处理) match the declared requirements (node/npm) and the install spec (npm package common-fetcher). No unrelated binaries or credentials are requested.
Instruction Scope
SKILL.md stays on-topic (CLI usage, Node API, config/ directory, openclaw.json integration). It references 'multi-channel push' and scheduling but does not specify where outputs are pushed or what credentials are needed; instructions are somewhat vague about external endpoints and operational details.
Install Mechanism
Install uses a public npm package name 'common-fetcher' (moderate risk). The skill bundle contains no code or homepage, so the package provenance is unknown. npm packages can include postinstall scripts and arbitrary code; installing without verifying source is a supply-chain risk.
Credentials
No environment variables or credentials are declared, which aligns with the minimal metadata. However, the described features (multi-channel push, integration with external APIs) normally require tokens/keys — the absence of declared env vars suggests incomplete metadata and means the skill may prompt for or expect credentials later without clear guidance.
Persistence & Privilege
always is false and no special system config paths are requested. The README suggests enabling/scheduling the skill via openclaw.json, which is normal. Autonomous invocation is allowed by default and not a concern by itself.
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 common-fetcher
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /common-fetcher 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
v1.0.0
Initial release - 207+ pre-configured sources (coal, realestate, AI) - 4 parsers validated (100% success rate) - <600ms performance for 30 articles - AI scoring, classification, and summarization - CLI and Node.js API support
元数据
Slug common-fetcher
版本 1.0.0
许可证
累计安装 1
当前安装数 1
历史版本数 1
常见问题

Common-Fetcher 是什么?

统一采集框架 - 支持 RSS/Web/API,207+ 采集源,AI 评分/分类/摘要. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 555 次。

如何安装 Common-Fetcher?

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

Common-Fetcher 是免费的吗?

是的,Common-Fetcher 完全免费(开源免费),可自由下载、安装和使用。

Common-Fetcher 支持哪些平台?

Common-Fetcher 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 Common-Fetcher?

由 luck(@lq707904686)开发并维护,当前版本 v1.0.0。

💬 留言讨论