KanseiLink MCP Server

MCP intelligence layer for discovering and orchestrating Japanese SaaS MCP tools.

KanseiLink helps AI agents find, evaluate, and combine Japanese SaaS services through the Model Context Protocol. It provides intent-based search, workflow recipes, change detection, and community-driven quality insights.

📝 Zenn記事: MCPサーバーが増えすぎて困ったので、MCPを整理するMCPサーバーを作った

Quick Start

npx @kansei-link/mcp-server

Or add to your MCP client config:

{
  "mcpServers": {
    "kansei-link": {
      "command": "npx",
      "args": ["@kansei-link/mcp-server"]
    }
  }
}

Tools

Tool	Description
search_services	Find Japanese SaaS MCPs by intent (FTS5 + category boost)
get_recipe	Get workflow patterns combining multiple services
find_combinations	Reverse lookup — find recipes containing a specific service
check_updates	Check recent changes and breaking updates for a service
report_outcome	Share your experience with auto PII masking
get_insights	Check community usage data and confidence scores

Categories

CRM, Project Management, Communication, Accounting, HR, E-commerce

Architecture

Agent ←→ KanseiLink MCP Server ←→ SQLite (local)
              ↓
         search_services   → FTS5 + intent→category mapping
         get_recipe        → Workflow pattern matching
         find_combinations → Reverse recipe lookup
         check_updates     → Changelog query
         report_outcome    → PII masking → outcomes table
         get_insights      → Aggregation + confidence scoring

Development

npm install
npm run build
npm run seed    # populate with Japanese SaaS MCP data
npm start       # start stdio server

Security

• PII auto-masking (Japanese kanji/katakana names, email, phone, IP)
• Agent identity anonymized
• See SECURITY.md for full policy

Links

• npm
• MCP Registry: io.github.kansei-link/kansei-mcp-server
• Zenn記事

License

MIT — Synapse Arrows PTE. LTD.