360 Secure SkillHub
360 安全技能中心

S³ Last 30 Days Research

该 Skill 先解析主题与查询类型,再通过多角度网页搜索和网页抓取,汇总 Reddit、X/Twitter 及一般网页中近 30 天的讨论,并按推荐、新闻、提示词或综合主题输出带来源的简报。

solomonneas
solomonneas
热度 22.72
最近更新 2026-03-20 02:51:48
分类:搜索依赖:联网浏览搜索应用场景:研究 / 社交媒体 / 摘要总结 / 报告撰写

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add s3-last30days
Metadata
name
last30days
description
Research any topic across Reddit, X/Twitter, and the web from the last 30 days. Surfaces current discussions, recommendations, trends, and news. Use when the user says /last30days or asks about recent trends, what people are saying about a topic, or current community sentiment.
argument-hint
cybersecurity trends, best AI coding tools, NVIDIA news, remote network admin jobs

last30days: Research Any Topic from the Last 30 Days

Research ANY topic across Reddit, X, and the broader web. Surface what people are actually discussing, recommending, and debating right now.

Parse User Intent

Before searching, extract:

  1. TOPIC: What they want to learn about
  2. QUERY_TYPE: One of:
    • RECOMMENDATIONS — "best X", "top X", "what X should I use" → wants a ranked list
    • NEWS — "what's happening with X", "X news", "latest on X" → current events
    • PROMPTING — "X prompts", "prompting for X" → techniques and copy-paste prompts
    • GENERAL — anything else → broad understanding

Display your parsing before searching:

Researching {TOPIC} across Reddit, X, and the web (last 30 days).
- QUERY_TYPE: {QUERY_TYPE}

Research Execution

Run 6-10 web searches using web_search with different query angles. Mix these source types:

Reddit searches (2-3 queries)

site:reddit.com {TOPIC} [relevant terms]
reddit {TOPIC} recommendations
reddit {TOPIC} discussion 2026

X/Twitter searches (2-3 queries)

site:x.com OR site:twitter.com {TOPIC}
{TOPIC} twitter thread
{TOPIC} trending

Web searches (2-4 queries)

Vary by QUERY_TYPE:

  • RECOMMENDATIONS: best {TOPIC} 2026, {TOPIC} comparison, top {TOPIC} list
  • NEWS: {TOPIC} news 2026, {TOPIC} announcement, {TOPIC} update
  • PROMPTING: {TOPIC} prompts examples, {TOPIC} techniques tips 2026
  • GENERAL: {TOPIC} 2026, {TOPIC} guide, {TOPIC} overview

Use freshness: "pm" (past month) on all searches when supported.

Deep-dive promising results

For the 3-5 most relevant search results, use web_fetch to get the full content. Prioritize:

  • Reddit threads with discussion
  • Blog posts with examples
  • News articles with details

Synthesis

After all research completes, synthesize into a single deliverable based on QUERY_TYPE:

If RECOMMENDATIONS

🏆 Most Mentioned (ranked by frequency across sources):

1. **[Name]** — [what it is, why people recommend it]
   Sources: [Reddit thread], [blog post], [X post]

2. **[Name]** — ...

Include honorable mentions. Note any controversial picks.

If NEWS

📰 Key Developments (last 30 days):

**[Headline]** — [date if known]
[2-3 sentence summary with source]

**[Headline]** — ...

Include community reaction where available.

If PROMPTING

🎯 Top Techniques (from community):

**[Technique Name]**
[Explanation + copy-paste example prompt]

**[Technique Name]**
...

If GENERAL

💡 What People Are Saying:

**Key Themes:**
1. [Theme] — [evidence from sources]
2. [Theme] — ...

**Notable Discussions:**
- [Summary of interesting thread/post with source]

**Consensus vs Debate:**
- Agreement: [what most agree on]
- Split: [where opinions diverge]

Output Rules

  • Cite sources with links where possible
  • Weight Reddit/X higher (engagement signals: upvotes, likes, replies)
  • Flag contradictions between sources
  • Be specific — names, numbers, dates. Not vague summaries.
  • Recency matters — prefer content from last 7 days over last 30
  • Keep total output under 800 words unless the user asks for deep detail