Last30days

/last30days Research Skill

Real-time intelligence engine: Find what's working RIGHT NOW, not last quarter.

Scans Reddit, X, and web for the last 30 days, identifies patterns, extracts community insights, and delivers actionable intelligence with copy-paste-ready prompts.

Mode

Detect from context or ask: "Quick pulse, full research, or strategic intelligence brief?"

Default: `standard` — use quick if they want a fast read. Use deep if they're making a business or product decision.

Why This vs ChatGPT?

Problem with "research [topic]": ChatGPT's training data is months/years old. It gives you general knowledge, not current signals.

Problem with Perplexity: Searches web but misses Reddit threads and X conversations where real practitioners share what's actually working.

This skill provides:

1. 30-day freshness filter - Only pulls recent content (not 2023 blog posts)
2. Multi-platform synthesis - Combines Reddit (detailed discussions), X (real-time signals), and web (articles) in one pass
3. Pattern detection - Highlights themes mentioned 3+ times across sources
4. Sentiment analysis - Shows community vibe (hype, skepticism, frustration)
5. Ready-to-use outputs - Copy-paste prompts and action ideas, not just summaries

You can replicate this by manually searching Reddit, X, and Brave Search with date filters, reading 30+ sources, identifying patterns, and synthesizing insights. Takes 2+ hours. This skill does it in 7 minutes.

When to Use

Perfect for:

• Trend discovery - "What's hot in AI agents right now?"
• Strategy validation - "What content marketing tactics are working in 2026?"
• Competitive intel - "What are developers saying about Cursor vs Copilot?"
• Product research - "What do users love/hate about Notion?"
• Prompt research - "What Claude prompting techniques are trending?"
• Community sentiment - "How do marketers feel about AI tools?"

Not ideal for:

• Historical research (use regular search)
• Academic/scientific papers (use Google Scholar)
• Non-English topics (limited coverage)
• Topics with zero online discussion

Required Setup

This skill orchestrates multiple tools. Verify you have:

# 1. Brave Search API (for web_search)
# Already configured in OpenClaw by default

# 2. Bird CLI (for X/Twitter search)
source ~/.openclaw/credentials/bird.env && bird search "test" -n 1
# If this fails, install bird CLI first

# 3. Reddit Insights (optional but recommended)
# If you have reddit-insights MCP server configured, skill will use it
# Otherwise falls back to Reddit web search via Brave

Quick verification:

/last30days --check-setup

Should return:

• ✅ Brave Search: Available
• ✅ Bird CLI: Available
• ✅ Reddit Insights: Available (or "Using web search fallback")

Workflow

Step 1: Web Search (Freshness Filter = Past Month)

web_search: "[topic] 2026" + freshness=pm
web_search: "[topic] strategies trends current"
web_search: "[topic] what's working"

Purpose: Get recent articles, blog posts, tools

Step 2: Reddit Search

If reddit-insights MCP configured:

reddit_search: "[topic] discussions techniques"
reddit_get_trends: "[subreddit]"

Otherwise:

web_search: "[topic] site:reddit.com" + freshness=pm
web_search: "[topic] reddit.com/r/[relevant_sub]"

Purpose: Find detailed discussions, practitioner insights, "what's actually working" threads

Step 3: X/Twitter Search

bird search "[topic]" -n 10
bird search "[topic] 2026" -n 10
bird search "[topic] best practices" -n 10

Purpose: Real-time signals, expert takes, trending threads

Step 4: Deep Dive on Top Sources (Optional)

For the 2-3 most relevant links:

web_fetch: [article URL]

Purpose: Extract specific tactics, quotes, data points

Step 5: Synthesize & Package

1. Identify patterns - What appears 3+ times across sources?
2. Extract key quotes - Most upvoted Reddit comments, retweeted takes
3. Assess sentiment - Hype, adoption, skepticism, frustration?
4. Create ready-to-use outputs - Prompts, action ideas, copy-paste tactics

Output Template

# 🔍 /last30days: [TOPIC]
*Research compiled: [DATE]*  
*Sources analyzed: [NUMBER] (Reddit threads, X posts, articles)*  
*Time period: Last 30 days*

---

## 🔥 Top Patterns Discovered

### 1. [Pattern Name]
**Mentioned: X times across [platforms]**

[Description of the pattern + why it matters]

**Key evidence:**
- Reddit (r/[sub]): "[Quote from highly upvoted comment]"
- X: "[Quote from popular thread]"
- Article ([Source]): "[Key insight]"

---

### 2. [Pattern Name]
[Continue same format...]

---

## 📊 Reddit Sentiment Breakdown

| Subreddit | Discussion Volume | Sentiment | Key Insight |
|-----------|-------------------|-----------|-------------|
| r/[sub] | [# threads] | 🟢 Positive / 🟡 Mixed / 🔴 Skeptical | [One-liner takeaway] |

**Top upvoted insights:**
1. "[Quote]" — u/[username] (+234 upvotes)
2. "[Quote]" — u/[username] (+189 upvotes)

---

## 🐦 X/Twitter Signal Analysis

**Trending themes:**
- [Theme 1] - [# mentions]
- [Theme 2] - [# mentions]

**Notable voices:**
- [@handle]: "[Key take]"
- [@handle]: "[Key take]"

**Engagement patterns:**
[What types of posts are getting traction?]

---

## 📈 Web Article Highlights

**Most shared articles:**
1. "[Article Title]" — [Source] — [Key insight]
2. "[Article Title]" — [Source] — [Key insight]

**Common recommendations across articles:**
- [Tactic 1]
- [Tactic 2]
- [Tactic 3]

---

## 🎯 Copy-Paste Prompt

**Based on current community best practices:**

[Ready-to-use prompt incorporating the patterns discovered]

Context: [Relevant context from research] Task: [Clear task] Style: [Tone/voice based on research] Constraints: [Any patterns to avoid based on research]

**Why this works:** [Brief explanation based on research findings]

---

## 💡 Action Ideas

**Immediate opportunities based on this research:**

1. **[Opportunity 1]**
   - What: [Specific action]
   - Why: [Evidence from research]
   - How: [Implementation steps]

2. **[Opportunity 2]**
   [Continue format...]

---

## 📌 Source List

**Reddit Threads:**
- [Thread title] - r/[sub] - [URL]

**X Threads:**
- [@handle] - [Tweet] - [URL]

**Articles:**
- [Title] - [Source] - [URL]

---

*Research complete. [X] sources analyzed in [Y] minutes.*

Real Examples

Example 1: Prompt Research

Query: /last30days Claude prompting best practices

Abbreviated Output:

# 🔍 /last30days: Claude Prompting Best Practices

## Top Patterns Discovered

### 1. XML Tags for Structure (12 mentions)
Reddit and X both emphasize using XML tags for complex prompts:
- Reddit: "XML tags changed my Claude workflow. <context> and <task> make responses 3× more accurate."
- X: "@anthropicAI's own docs now recommend XML. It's the meta."

### 2. Examples Over Instructions (9 mentions)  
"Show, don't tell" — Provide 2-3 examples instead of long instructions.

### 3. Chain of Thought Explicit (7 mentions)
Add "Think step-by-step before answering" dramatically improves reasoning.

## Copy-Paste Prompt

<context>
[Your context here]
</context>

<task>
[Your task here]
</task>

<examples>
Example 1: [Show desired output style]
Example 2: [Show edge case handling]
</examples>

Think step-by-step before providing your final answer.

Example 2: Competitive Intel

Query: /last30days Notion vs Obsidian 2026

Abbreviated Output:

## Top Patterns