/research

Deep research before PRD generation. Produces a structured research.md with competitive analysis, user pain points, SEO/ASO keywords, naming/domain options, and market sizing.

MCP Tools (use if available)

If MCP tools are available, prefer them over CLI:

• kb_search(query, n_results) — search knowledge base for related docs
• web_search(query, engines, include_raw_content) — SearXNG web search with engine routing
• session_search(query, project) — find how similar research was done before
• project_info(name) — check project details

MCP web_search supports engine override: engines="reddit", engines="youtube", etc. If MCP tools are not available, use Claude WebSearch/WebFetch as fallback.

Search Strategy: Hybrid (SearXNG + Claude WebSearch)

Use both backends together. Each has strengths:

SearXNG Availability

If MCP web_search is available, use it (it wraps SearXNG). Otherwise, check local availability:

curl -sf http://localhost:8013/health && echo "searxng_ok" || echo "searxng_down"

If unavailable — use Claude WebSearch for everything.

SearXNG via curl (when MCP not available)

# General query
curl -s -X POST 'http://localhost:8013/search' \
  -H 'Content-Type: application/json' \
  -d '{"query": "<query>", "max_results": 10, "include_raw_content": true}'

# Force Reddit only
curl -s -X POST 'http://localhost:8013/search' \
  -H 'Content-Type: application/json' \
  -d '{"query": "<query>", "max_results": 10, "engines": "reddit", "include_raw_content": true}'

# YouTube transcript
curl -s -X POST 'http://localhost:8013/transcript' \
  -H 'Content-Type: application/json' \
  -d '{"video_id": "<VIDEO_ID_or_URL>", "max_length": 5000}'

Steps

1. Parse the idea from $ARGUMENTS. If empty, ask the user what idea they want to research.

2. Ask product type — ask the user:

   • Web app / Mobile (iOS) / Mobile (Android) / Desktop / CLI / API
   • This determines which research sections apply (ASO for mobile, SEO for web, etc.)

3. Search knowledge base for existing related content:

   • If MCP kb_search available: kb_search(query="<idea keywords>", n_results=5)
   • Otherwise: Grep for keywords in .md files
   • Check if research.md or prd.md already exist for this idea.

4. Competitive analysis — use Claude WebSearch (primary) + SearXNG (scraping):

   • "<idea> competitors alternatives 2026" — broad discovery
   • "<idea> app review pricing" — pricing data
   • SearXNG include_raw_content=true: scrape competitor URLs for detailed pricing
   • SearXNG/MCP engines: reddit: "<idea> vs" — user opinions
   • "site:producthunt.com <idea>" — Product Hunt launches
   • "site:g2.com <idea>" or "site:capterra.com <idea>" — B2B reviews
   • "site:crunchbase.com <competitor>" — funding, team size
   • For each competitor extract: name, URL, pricing, key features, weaknesses

5. User pain points — use SearXNG reddit (primary) + YouTube + Claude WebSearch:

   • SearXNG/MCP engines: reddit: "<problem>" — Reddit via PullPush API
   • SearXNG/MCP engines: youtube: "<problem> review" — video reviews
   • SearXNG /transcript: extract subtitles from top 2-3 YouTube videos
   • "site:news.ycombinator.com <problem>" — Hacker News opinions
   • Claude WebSearch: "<problem> frustrating OR annoying" — broader sweep
   • Synthesis: top 5 pain points with quotes and source URLs

6. SEO / ASO analysis (depends on product type from step 2):

   For web apps:

   • "<competitor> SEO keywords ranking" — competitor keywords
   • "<problem domain> search volume trends 2026" — demand signals
   • SearXNG include_raw_content: scrape competitor pages for meta tags
   • Result: keyword table (keyword, intent, competition, relevance)

   For mobile apps:

   • "<category> App Store top apps keywords 2026" — category landscape
   • "site:reddit.com <competitor app> review" — user complaints
   • Result: ASO keywords, competitor ratings, common complaints

7. Naming, domains, and company registration:

   • Generate 5-7 name candidates

   Domain availability (RDAP — free, no auth):

   # 404 = available, 200 = registered
   curl -s -o /dev/null -w "%{http_code}" "https://rdap.org/domain/<name>.com"
   

   • Check .com, .app, .io, .dev, .ai for each candidate (run in parallel)

   Trademark check:

   • "<name> trademark" — basic conflict check
   • "site:trademarkia.com <name>" — US trademarks

   Company name conflicts:

   • "site:opencorporates.com <name>" — 240M+ companies worldwide
   • "<name> LLC OR Inc OR Corp" — existing businesses

8. Market sizing (TAM/SAM/SOM) — use Claude WebSearch (primary):

   • "<market> market size 2025 2026 report" — synthesizes numbers
   • "<market> growth rate CAGR billion" — growth projections
   • Extrapolation: TAM → SAM → SOM (Year 1)

9. Write research.md — write to 3-opportunities/<project-name>/research.md if in solopreneur KB, otherwise to docs/research.md in the current project. Create the directory if needed.

10. Output summary:

    • Key findings (3-5 bullets)
    • Recommendation: GO / NO-GO / PIVOT with brief reasoning
    • Path to generated research.md
    • Suggested next step: /validate <idea>

research.md Format

---
type: research
status: draft
title: "Deep Research — <Idea Name>"
created: <today>
tags: [<project>, research, competitive-analysis]
product_type: web|ios|android|desktop|cli|api
---

# Deep Research: <Idea Name>

## Executive Summary
<!-- 3-4 sentences: problem, market, competitors, recommendation -->

## 1. Competitive Landscape

| Competitor | URL | Pricing | Key Features | Weaknesses |
|-----------|-----|---------|-------------|------------|
| ... | ... | ... | ... | ... |

### Gap Analysis
<!-- What nobody does. Where our advantage is. -->

## 2. User Pain Points

| Pain Point | Source | URL | Sentiment |
|-----------|--------|-----|-----------|
| ... | Reddit r/... | ... | negative |

### Top Insights
<!-- 3-5 key insights -->

## 3. SEO / ASO Analysis

| Keyword | Intent | Competition | Relevance |
|---------|--------|------------|-----------|
| ... | commercial | medium | primary |

## 4. Naming & Domains

| Name | .com | .app | .io | .ai | Trademark | Notes |
|------|------|------|-----|-----|-----------|-------|
| ... | avail | taken | avail | avail | clean | ... |

### Recommended Name: **<name>**

## 5. Market Size

- **TAM:** $X — <source>
- **SAM:** $X — <reasoning>
- **SOM (Year 1):** $X — <assumptions>

## 6. Recommendation

**Verdict:** GO / NO-GO / PIVOT

## Sources

1. [Title](url) — brief description

Notes

• Always use kebab-case for project directory names
• If research.md already exists, ask before overwriting
• Run SearXNG and Claude WebSearch queries in parallel when independent