---
description: Riset topik/teknologi dari web — search, baca docs, compile findings
---

# Research (Enhanced D1337 Edition)

## Pre-Read Skills
```
view_file: ~/.gemini/skills/mindset/SKILL.md
view_file: ~/.gemini/skills/deep-recon/SKILL.md
view_file: ~/.gemini/skills/get-competitor-skill-and-method/SKILL.md
view_file: ~/.gemini/skills/underground-tools/SKILL.md
```

## Research Classification

Before starting, classify research type:

| Type | Depth | Min Sources | Primary Tools |
|------|-------|-------------|---------------|
| **TECH** | Library/framework/tool research | 3+ | `context7` → `search_web` → `read_url_content` |
| **CVE/VULN** | Vulnerability/exploit research | 5+ | `search_web` → `read_url_content` → NVD/Wordfence/Patchstack |
| **OSINT** | Target intelligence gathering | 3+ | `search_web` → SecurityTrails API → IntelX API |
| **UNDERGROUND** | Darknet tools/services/markets | 3+ | `search_web` → IntelX → forum monitoring → sample analysis |
| **MARKET** | Competitor/market analysis | 5+ | `search_web` → `firecrawl_search` → `read_url_content` |
| **STRATEGY** | Business/revenue/growth research | 5+ | `search_web` → `firecrawl_search` → `read_url_content` |

## Steps

// turbo-all

### 1. Define Scope
Ask Ketua (or infer from context) what topic to research.
Classify research type from table above.

### 2. Multi-Angle Web Search
Use `search_web` with **minimum 3 different query angles** for the topic.

**Query Strategy:**
```
Angle 1: Direct — "[topic] 2025 2026 latest"
Angle 2: Technical — "[topic] tutorial guide documentation"
Angle 3: Comparative — "[topic] vs alternatives comparison"
Angle 4: GitHub/Code — "site:github.com [topic]" OR "[topic] open source repository"
Angle 5: Problems — "[topic] issues problems limitations"
```

Focus on:
- Latest results (2025-2026 ONLY — older = deprecated intel)
- GitHub repos (recently updated, high stars)
- Official documentation (always primary source)
- Research papers / blog posts (technical depth)
- Hacker News / Reddit discussions (real-world experiences)

### 3. Context7 Library Docs (for TECH type)
If researching a library/framework:
```
1. mcp_context7_resolve-library-id → get library ID
2. mcp_context7_query-docs → get latest API, examples, patterns
3. Use this as GROUND TRUTH over blog posts
```

### 4. Deep Read
For the **top 3-5** most promising results:
- Use `read_url_content` for articles, docs, blog posts
- Use `browser_subagent` for JS-heavy pages, interactive content
- Use `firecrawl_scrape` for complex pages (LAST RESORT — paid)

**Extract from each source:**
- Key claims/findings
- Code examples if applicable
- Version numbers / dates
- Author credibility assessment

### 5. API Intelligence (for OSINT/CVE type)
If research type is OSINT or CVE:

**SecurityTrails** (domain/infra research):
```bash
curl -s "https://api.securitytrails.com/v1/domain/TARGET/subdomains" -H "apikey: KEY"
curl -s "https://api.securitytrails.com/v1/history/TARGET/dns/a" -H "apikey: KEY"
```

**IntelX** (leak/darknet research):
```bash
curl -s -X POST "https://2.intelx.io/intelligent/search" \
  -H "x-key: KEY" -H "Content-Type: application/json" \
  -d '{"term":"TARGET","maxresults":100,"sort":2,"timeout":5}'
```

**Shodan** (infrastructure research):
```
search_web "site:shodan.io TARGET"
```

### 6. Cross-Reference & Validate
Compare findings from multiple sources:
- Flag contradictions between sources
- Prefer: Official docs > GitHub releases > Blog posts > Forum answers
- Check dates — anything >12 months old needs verification
- Identify consensus vs outlier opinions
- Apply `mindset` skill: intelligence synthesis, pattern recognition

**Validation checklist:**
- [ ] Claims backed by code/evidence?
- [ ] Multiple independent sources confirm?
- [ ] Date/version still current?
- [ ] Author has track record?

### 7. Compile Report
Create an artifact with:

```markdown
# Research Report: [TOPIC]

## Executive Summary
- 3-5 bullet points, key takeaways

## Detailed Findings
### Source 1: [Title] (date, credibility)
[Findings]
### Source 2: ...

## Code Examples (if applicable)
[Working code, not snippets]

## Comparison Matrix (if applicable)
| Feature | Option A | Option B | Option C |
|---------|----------|----------|----------|

## Recommendations
- Primary recommendation with rationale
- Alternative approaches
- Risk factors

## Sources
1. [Title](URL) — accessed [date]
2. ...
```

### 8. Quality Gate
BEFORE delivering output, self-check:

- "Bisa dapet dari ChatGPT gratisan?" → YA → **REDO LEBIH DALAM**
- "Ada specific actionable takeaways?" → TIDAK → **TAMBAH DETAIL**
- "Udah cross-reference minimal 3 sources?" → TIDAK → **RISET LAGI**
- "Info masih current (2025-2026)?" → TIDAK → **UPDATE**
- "Code examples working/tested?" → TIDAK → **VERIFY**

### 9. Present to Ketua
Share the artifact and ask if deeper dive needed on any specific area.
Offer follow-up options:
- "Mau gw deep dive di [specific area]?"
- "Mau gw buatin PoC/implementation?"
- "Ada angle lain yang mau gw riset?"