What is KanseiLink?

KanseiLink is an MCP server that gives AI agents current, verified data on 900+ SaaS services — before they waste tokens on trial-and-error with stale training data.

The Problem

AI agents' training data for SaaS APIs is 6-12 months old. In that time:

• Endpoints change — deprecated paths return 404
• Auth flows break — OAuth scopes get renamed, token lifetimes shorten
• Rate limits shift — what worked at 100 req/min now throttles at 30

The result: agents spend 15,000-25,000 tokens on web searches, doc scraping, and failed API calls just to figure out how to connect to a single service.

The Solution

KanseiLink replaces that trial-and-error loop with a single tool call:

lookup({ service_id: "freee" })

Returns: auth setup, endpoints, rate limits, pitfalls other agents hit, and workarounds that actually work — in ~800 tokens.

That's a 91-96% token savings over web search.

How It Works

KanseiLink is built on an ant colony model. Every agent that uses a service reports back:

1. Did it work? Success/failure rates build a real-time reliability map
2. What went wrong? Error patterns and workarounds are shared across the colony
3. Trust scores update automatically — if a service starts failing, agents see the warning immediately

This creates a self-correcting feedback loop: the more agents use KanseiLink, the more accurate it becomes.

5 Tools, That's All

Tool	Purpose	Who
search_services	Find the right service	All agents
lookup	Get tips, detail, insights, recipes, history	All agents
report	Report outcomes, submit feedback	All agents
inspect	Colony health, anomaly inspection	Operators
analyze	Token savings, cost audit, AEO reports	Operators

Most agents only need the first 3.

Quick Install

Claude Desktop

npx @kansei-link/mcp-server

Cursor / Windsurf

npx @kansei-link/mcp-server

claude_desktop_config.json

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

Quickstart — Get up and running in 2 minutes