Token导航 LogoToken导航

Agent Skill · 研究检索

yqg-ai-news-digest(yqgAI 新闻文摘)

yqg-ai-news-digest 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

yqg-ai-news-digest 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为调用外部服务;许可证为 MIT-0,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 21,803。

总安装 4,749周安装 204GitHub Stars 公开资料未说明下载 1,665核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
3.0.3

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:yqg-ai-news-digest(yqgAI 新闻文摘)
来源仓库:https://github.com/yqghlx/yqg-ai-news-digest
安装命令:
openclaw skills install yqg-ai-news-digest
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install yqg-ai-news-digest

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

聚合人工智能领域新闻并生成结构化报告的助手。自动收集、去重、分类最新 AI 发展动态。覆盖技术、应用与市场多个维度信息整合。yqg-ai-news-digest 属于研究检索类 Skill,可作为该场景下的辅助能力补充。适合研究人员与从业者跟踪行业趋势使用。提供每日简报与专题分析报告两种输出形式。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install yqg-ai-news-digest;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

yqg-ai-news-digest 怎么安装?

在终端执行 openclaw skills install yqg-ai-news-digest。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

yqg-ai-news-digest 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

yqg-ai-news-digest 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

yqg-ai-news-digest 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

yqg-ai-news-digest 用的人多吗?

公开数据是累计安装 4,749 次,近一周安装 204 次,下载 1,665 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

yqg-ai-news-digest 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
yqg-ai-news-digest

yqgAI 新闻文摘 · 本页

4,749—调用外部服务MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
ai-news-digest
description
AI industry news aggregator and structured report generator. Automatically collects, deduplicates, ranks, and categorizes the latest AI developments from multiple sources into professional daily or weekly digests. Supports topic deep-dives, competitor tracking, trend analysis, bilingual output, and multi-format delivery (chat, file, Feishu doc). Zero dependencies — pure prompt-driven. Use when user asks for AI news, tech digest, industry update, competitor analysis, or weekly AI roundup.

AI News Digest | AI 科技新闻日报

Zero-config AI news aggregation. No scripts, no APIs, no setup. From raw search results → polished, insight-ready report in one conversation.

Why This Skill Wins

Problem with other toolsHow we solve it
Need RSS feeds, Python scripts, cron jobsPure prompt-driven, zero setup
Dumps raw links at youStructured categories with summaries
No deduplicationSmart merge: same event from 5 sources → 1 entry
No rankingSignificance-based prioritization (critical → minor)
One-size output4 depth levels + custom topics
English-onlyBilingual (Chinese + English) by default

Activation Triggers

  • "AI 新闻" / "科技日报" / "AI 动态"
  • "What's new in AI today?"
  • "追踪大模型/AI Agent/AI编程赛道"
  • "分析 OpenAI/Google/Anthropic 最新动态"
  • "AI 周报" / "weekly AI roundup"
  • "AI 行业趋势"

Workflow: Daily Digest (Default Mode)

Phase 1 — Discovery (8 parallel searches)

Use web_search with count: 8 for each. Run all searches before proceeding.

#QueryPurpose
1"AI breaking news today {current_year}"General breaking
2"LLM model release launch {current_year}"New model releases
3"AI startup funding investment {current_year}"Funding rounds
4"人工智能 最新消息 今天"Chinese-language news
5"AI product launch announcement"Product updates
6"AI research breakthrough paper"Academic/research
7"OpenAI Google Anthropic Meta AI update"Big tech moves
8"AI regulation policy government"Policy changes

Focus mode — If user specified a topic, replace queries 5-8 with topic-specific searches:

TopicReplacement queries
AI Agents"AI agent framework launch", "autonomous agent 2026", "AI agent startup", "MCP tool use agent"
AI Coding"AI coding assistant update", "Copilot Cursor Windsurf Claude Code", "AI developer tool", "code generation model"
AI Video/Image"AI video generation Sora Runway Kling", "AI image Midjourney DALL-E", "AI creative tool", "generative AI art"
AI Hardware"AI chip GPU NVIDIA AMD", "AI inference hardware", "edge AI chip", "AI data center"
Robotics"humanoid robot", "AI robotics startup", "autonomous vehicle AI", "embodied AI"

Phase 2 — Extraction

For each result that looks relevant:

  1. Use web_fetch with maxChars: 3000
  2. Extract: headline, 2-3 sentence summary, source name, date, significance level
  3. Skip if: paywalled with no content, clearly outdated (>7 days), duplicate of already-extracted article, product was already released before the report period (verify launch date before listing as "new release")

Date filtering rules (STRICT — NO EXCEPTIONS):

  • 默认只收录当天(today)的新闻。用户说"今天"就是今天,说"3月31日"就是3月31日。
  • 不要包含昨天、前天、上周的新闻。用户如果需要最近几天/几周的新闻,会明确说"最近3天"、"本周"、"最近一周"等。
  • 每条新闻发布前必须确认发布日期是今天,不确定的标注"[日期待确认]",不得放入头条。
  • 空板块显示"暂无相关动态",绝不用旧新闻填充。

Source credibility tiers (prefer higher-tier when multiple sources cover same story):

  • 🥇 Tier 1: Official blogs (openai.com, blog.google, anthropic.com, ai.meta.com), arxiv.org
  • 🥈 Tier 2: techcrunch.com, theverge.com, arstechnica.com, 36kr.com, theinformation.com
  • 🥉 Tier 3: General news, blogs, social media posts

Phase 3 — Deduplication & Ranking

Deduplication rules:

  1. Same event, multiple sources → keep most detailed article, note other sources as "also reported by: ..."
  2. Follow-up coverage of ongoing story → merge into one entry, update with latest development
  3. Translation of same article (EN/CN) → keep in user's preferred language

Significance ranking:

LevelCriteriaExamples
🔴 CriticalIndustry-changing eventNew GPT release, major AI safety incident, billion-dollar deal, government ban
🟡 NotableSignificant but expectedProduct update, $10M+ funding, benchmark record, partnership
🟢 RoutineNormal activityMinor feature, opinion piece, routine update, rumor

Phase 4 — Output

Depth levels (user can specify, default: standard):

LevelContentBest for
BriefHeadlines + one-liners onlyQuick glance, mobile reading
StandardHeadlines + 2-3 sentence summaries + linksDefault daily use
DeepFull summaries + context + analysisResearch, decision-making
Executive3-5 bullet points onlyBusy leaders, standup prep

Standard depth output format:

# 🤖 AI 科技日报 | {YYYY-MM-DD 周X}

---

## 🔴 今日头条

### {Headline}
{2-3 sentences: what happened, why it matters, what's next}
🔗 [{Source}]({url}) | 也被报道: {Other Source 1}, {Other Source 2}

---

## 📰 行业动态
| 事件 | 要点 | 来源 |
|------|------|------|
| {Title} | {One-line summary} | [🔗]({url}) |

## 💰 投融资
- **{Company}** — {金额} {轮次} (领投: {Investor}) → 用途: {use of funds}
  🔗 [{Source}]({url})

## 🛠️ 技术突破
- **{What}** — {Why it matters, technical detail}
  🔗 [{Source}]({url})

## 📱 产品发布
- **{Product}** ({Company}) — {Key feature or change}
  🔗 [{Source}]({url})

## 📜 政策法规
- {What changed} — {Who is affected}
  🔗 [{Source}]({url})

---

💡 **趋势洞察**: {1-2 sentences connecting today's stories to broader patterns}
📋 **明日关注**: {Upcoming events, expected announcements, developing stories}

Advanced Modes

Mode: Topic Deep-Dive

User: "深入分析 {topic}" / "deep dive into {topic}"

  1. Run 10+ focused searches from different angles
  2. Fetch top 15 articles
  3. Output a research brief:
# 📊 {Topic} 深度分析 | {Date}

## 背景概述
{What led to this topic becoming relevant — 3-4 sentences}

## 最新进展
{Timeline of recent events, newest first}

## 核心玩家
| 公司/团队 | 动作 | 策略方向 |
|-----------|------|----------|
| {Name} | {What they did} | {Where they're heading} |

## 技术分析
{Deeper look at the technology, 1-2 paragraphs}

## 市场影响
{How this affects the industry, investment landscape, etc.}

## 未来展望
{What to expect in next 3-6 months}

## 完整来源
1. [{Title}]({url}) — {Source}
2. ...

Mode: Competitor Radar

User: "分析 {Company} 的最新 AI 动态"

# 🔍 {Company} 竞品雷达 | {Date}

## 近30天动态
| 日期 | 事件 | 类型 |
|------|------|------|
| {Date} | {What happened} | 产品/技术/人事/战略 |

## 产品路线分析
{Where their product is heading based on recent moves}

## 关键人才变动
{Notable hires or departures}

## 研究/专利
{Recent papers, patents, or open-source releases}

## 对我们的启示
{Actionable insights for the user's own strategy}

Mode: Weekly Roundup

User: "AI 周报" / "weekly AI roundup"

Daily format + these extra sections:

## 📈 本周趋势
1. **{Trend 1}**: {Evidence from multiple stories}
2. **{Trend 2}**: {Evidence}

## 📊 数据亮点
- 融资总额: ${amount} across {n} rounds
- 新模型发布: {n} 个
- 重要产品: {highlights}

## 📅 下周看点
- {Upcoming event 1}
- {Expected announcement}
- {Ongoing story to follow}

Quality Guarantee Checklist

Before delivering output, verify EVERY item:

  • [ ] Every claim has a clickable source link
  • [ ] No duplicate entries for the same event
  • [ ] All sections present (use "暂无相关动态" for empty ones)
  • [ ] Critical events are in the 头条 section
  • [ ] No content older than 7 days
  • [ ] Objectivity: facts only, no editorializing
  • [ ] Language matches user preference (Chinese default, English on request)
  • [ ] Proper nouns kept in English (GPT-4o, not "GPT-4哦")

Error Handling

SituationAction
Search API fails for a querySkip, continue with remaining searches
Article fetch returns emptyUse search snippet, mark "[摘要来自搜索结果]"
All searches failReturn: "⚠️ 搜索服务暂时不可用,请稍后再试"
No news in a categoryShow "暂无相关动态" — never omit the section
Source is paywalledUse snippet, mark "[付费文章,仅摘要]"

Delivery Options

TargetHow
Chat (default)Markdown inline
Feishu docfeishu_doc tool → create → write
FileSave to user-specified path
EnglishUser says "in English" → all labels switch to English

User Satisfaction Tips

  • Speed matters: Run searches in parallel (batch all web_search calls)
  • Be concise: Don't pad with filler — every sentence should inform
  • Add value: Don't just list news; connect dots in 趋势洞察
  • Respect time: If user asks for "简报", don't give them a wall of text
  • Follow up: If a story seems incomplete, note it in 明日关注

平台分布

OpenClaw

87.6%
按下载量换算1,459

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills