mcp-health is an MCP server that analyzes your MCP setup — measuring the token cost of every installed server and tool, finding duplicates, grading efficiency, and generating optimization reports.

Quick Start

npx mcp-health

Then ask your AI assistant:

"Run an MCP health check on my setup"
"How many tokens are my MCP servers using?"
"Generate an MCP health report"

Why?

Every MCP tool costs 550–1,400 tokens just for its schema. A server with 50 tools can eat 30,000+ tokens before you type anything. Most developers have no idea how much context window their MCP servers are consuming.

Real examples from the community:

Adding GitHub MCP increased one user's token usage from 34K to 80K (source)
One developer found 67,000 tokens consumed by just 4 servers before any conversation (source)
Context7 (240K weekly downloads) has tool descriptions costing 3x more tokens than necessary (source)

Tools

`analyze_servers`

Scan your MCP config and measure the token cost of every installed server. Auto-detects config files.

"Which of my MCP servers costs the most tokens?"

`analyze_tools`

Deep-dive into a specific server — per-tool token costs, grades, bloated descriptions, and optimization suggestions.

"Break down the token cost of my GitHub MCP server"

`find_duplicates`

Find duplicate tool names across your installed servers. Redundant tools waste context for no benefit.

"Do any of my MCP servers have overlapping tools?"

`count_tokens`

Count tokens in any text. Useful for estimating context usage.

"How many tokens is this prompt?"

`generate_report`

Full markdown health report: server grades (A-F), token costs, duplicates, and actionable recommendations.

"Generate a full MCP health report"

Grading System

Per-Tool Grades

Grade	Tokens	Meaning
A	≤100	Excellent — minimal context cost
B	≤300	Good — reasonable schema size
C	≤600	Fair — could be optimized
D	≤1000	Poor — bloated schema
F	>1000	Failing — severely impacts context window

Per-Server Grades

Grade	Total Tokens	Meaning
A	≤500	Lean and efficient
B	≤1500	Reasonable
C	≤3000	Getting heavy
D	≤6000	Significant context cost
F	>6000	Major context window drain

Installation

Claude Desktop / Claude Code

{
  "mcpServers": {
    "mcp-health": {
      "command": "npx",
      "args": ["-y", "mcp-health"]
    }
  }
}

Cursor

Add to .cursor/mcp.json with the same format.

Auto-Detection

mcp-health automatically finds your MCP config in these locations:

.mcp.json (current directory)
Claude Desktop config (Windows/macOS/Linux)
.cursor/mcp.json

Or pass a custom path to any tool.

License

MIT

Want to build your own MCP server? Check out our MCP Server Development Kit — includes TypeScript templates, market research, niche analysis, and monetization strategies.