korean-keyword-mcp
# korean-keyword-mcp
MCP server for Korean keyword niche analysis. The **only** MCP server that exposes Naver SearchAd API data (CPC competition, click rates) for keyword research.
## Why This Exists
| Feature | korean-keyword-mcp | Other Naver MCP servers |
|---|---|---|
| SearchAd API (CPC, competition index) | **YES** | NO |
| Niche scoring (0-100, A-F grade) | **YES** | NO |
| Blog competition deep analysis | **YES** | Basic search only |
| Keyword expansion + scoring | **YES** | NO |
| Batch comparison analysis | **YES** | NO |
| Trending keyword discovery | **YES** | NO |
## Tools (7)
| Tool | Description |
|------|-------------|
| `keyword_expand` | Expand seed keyword → 50+ related keywords with Quick Score + top N Full Score |
| `niche_score` | Single keyword full niche analysis (0-100 score + A-F grade) |
| `search_volume` | Naver SearchAd monthly search volume (PC/mobile split, competition index) |
| `trend` | Naver DataLab 12-month search trend (monthly relative values 0-100) |
| `blog_competition` | Blog competition analysis (total results + top 10 posts) |
| `batch_analyze` | Batch niche analysis for up to 10 keywords with sorted comparison |
| `trending_discover` | Discover rising-trend keywords from seed keyword's related keywords |
## Scoring Algorithm
**Full Niche Score (100 points)**
| Component | Weight | Source |
|-----------|--------|--------|
| Volume | 20 | SearchAd API — sweet spot: 1K-30K searches |
| Competition | 30 | Blog total results + blogger diversity |
| Freshness | 15 | Average post age (older = less competition) |
| Trend | 20 | 12-month linear regression slope |
| Efficiency | 15 | Search volume / blog post ratio |
**Grades**: A (75+), B (60+), C (45+), D (30+), F (<30)
## Setup
### 1. Get Naver API Keys
You need two sets of API credentials:
**Naver SearchAd API** (for search volume, CPC, competition):
1. Go to [Naver SearchAd](https://searchad.naver.com)
2. Create an account → Tools → API License
3. Note your Customer ID, API Key, and Secret Key
**Naver Developer API** (for DataLab trends, blog search):
1. Go to [Naver Developers](https://developers.naver.com)
2. Register application → Select "Search" and "DataLab" APIs
3. Note your Client ID and Client Secret
### 2. Configure Claude Desktop
Add to your Claude Desktop config (`claude_desktop_config.json`):
```json
{
"mcpServers": {
"korean-keyword": {
"command": "npx",
"args": ["-y", "korean-keyword-mcp"],
"env": {
"NAVER_SEARCHAD_CUSTOMER_ID": "your-customer-id",
"NAVER_SEARCHAD_API_KEY": "your-api-key",
"NAVER_SEARCHAD_SECRET_KEY": "your-secret-key",
"NAVER_CLIENT_ID": "your-client-id",
"NAVER_CLIENT_SECRET": "your-client-secret"
}
}
}
}
```
### 3. Verify
Restart Claude Desktop. You should see "korean-keyword" in the MCP servers list with 7 tools available.
## Example Usage
Once connected, you can ask Claude:
- "Analyze the niche potential of '캠핑의자'" → uses `niche_score`
- "Find niche keywords related to '다이어트'" → uses `keyword_expand`
- "Compare these keywords: 캠핑의자, 캠핑테이블, 캠핑조명" → uses `batch_analyze`
- "What keywords related to '캠핑' are trending up?" → uses `trending_discover`
- "How much search volume does '에어프라이어' get?" → uses `search_volume`
## Environment Variables
| Variable | Required | Description |
|----------|----------|-------------|
| `NAVER_SEARCHAD_CUSTOMER_ID` | Yes | SearchAd API customer ID |
| `NAVER_SEARCHAD_API_KEY` | Yes | SearchAd API key |
| `NAVER_SEARCHAD_SECRET_KEY` | Yes | SearchAd API secret |
| `NAVER_CLIENT_ID` | Yes | Naver Developer client ID |
| `NAVER_CLIENT_SECRET` | Yes | Naver Developer client secret |
## License
MIT
TDQS
Scored across 7 tools
Several tools compose or reuse the same underlying operations (keyword_expand includes niche scoring; trending_discover expands and filters; batch_analyze repeats niche_score for many keywords), so boundaries are not always crisp. However, descriptions clarify the intended use case well enough for most selections.
Names use a consistent snake_case style and a clear keyword/niche domain prefix, but the grammatical pattern is mixed: keyword_expand and batch_analyze are noun+verb, while search_volume, blog_competition, and trend are noun phrases. There is no uniform verb-first or verb-last convention across the set.
Seven tools cover the core keyword research workflow—expansion, individual metrics, scoring, batch comparison, and trend discovery—without redundancy or bloat. The count feels appropriate for the server's focused purpose.
The toolset covers expansion, volume, trend, competition, scoring, batch analysis, and rising-trend discovery, which handles the main niche research workflow. Minor gaps remain, such as no raw multi-keyword volume comparison or direct trend comparison outside the scored outputs.