kd-scout
# kd-scout
Zero-dependency MCP server for keyword research arithmetic: difficulty estimation,
opportunity scoring, and structured content briefs. Runs anywhere Python 3.10+ runs -
no packages, no API keys required for the deterministic mode.
## Tools
| Tool | Input | Output |
|---|---|---|
| `keyword_difficulty` | `keywords[]` | per-keyword difficulty 0-100 |
| `opportunity_score` | `difficulty`, `monthly_volume` | blended 0-100 score |
| `content_brief` | `topic` | H2 outline, word-count target, FAQs, schema type |
## Install
```bash
pip install kd-scout
kd-scout # stdio MCP server
```
Or point any MCP client at the module directly:
```json
{"command": "kd-scout"}
```
## Production use
The scoring model is derived from the editorial pipeline behind
[Groundwork's money & tools research](https://gworky.com/money), where it screens
hundreds of candidate keywords weekly.
## License
MIT
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose in the keyword research workflow: difficulty estimation, opportunity scoring, and content brief generation. There is meaningful sequential dependency, but no functional overlap that would confuse an agent.
All tool names follow the same lowercase snake_case pattern and describe their output: keyword_difficulty, opportunity_score, content_brief. The naming convention is predictable and easy to extend.
Three tools is perfectly scoped for this server's narrow purpose. Each tool fills a distinct step in the pipeline without redundancy or unnecessary bloat.
The set covers the core lifecycle from raw signals through difficulty and opportunity scoring to a content brief, leaving no obvious workflow dead end. It lacks built-in SERP or volume retrieval, but those appear to be external inputs rather than gaps in the server's stated scope.