mcp-checkup
About
Analyze your MCP setup: measure token cost per tool, find bloated schemas, detect duplicates, and get optimization tips to reclaim your context window.
Details
- Author
- yifanyifan897645
- Categories
- Developer Tools, Infrastructure, Other
Jump to
Setup
Install mcp-checkup in your MCP client (Claude Desktop, Cursor, Windsurf, and others).
Repository: https://github.com/yifanyifan897645/mcp-checkup
Follow the installation instructions in the repository README, then restart your MCP client.
Your MCP servers are eating your context window. Find out how much.
mcp-checkupis 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.
"Run an MCP health check on my setup" "How many tokens are my MCP servers using?" "Generate an MCP health report"
Every MCP tool costs550–1,400 tokensjust for its schema. A server with 50 tools can eat30,000+ tokensbefore you type anything. Most developers have no idea how much context window their MCP servers are consuming.
- 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)
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?"
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 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 in any text. Useful for estimating context usage.
"How many tokens is this prompt?"
Full markdown health report: server grades (A-F), token costs, duplicates, and actionable recommendations.
"Generate a full MCP health report"
{ "mcpServers": { "mcp-checkup": { "command": "npx", "args": ["-y", "mcp-checkup"] } } }
Add to.cursor/mcp.jsonwith the same format.
mcp-checkup automatically finds your MCP config in these locations:
- .mcp.json(current directory)
- Claude Desktop config (Windows/macOS/Linux)
- .cursor/mcp.json
Use mcp-checkup firstto find the problem, then decide how to fix it.
- webcheck-mcp— Website health analysis
- git-summary-mcp— Git repository intelligence
- mcp-checkup— MCP setup health analyzer
- dev-utils-mcp— Developer utilities
- codescan-mcp— Codebase health scanner
- deadlink-checker-mcp— Dead link detector
This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.
AI-powered live runtime debugging with Lightrun production context.
MCP server for .NET runtime diagnostics — memory, GC, threads, and more.
MCP server for .NET runtime diagnostics — memory, GC, threads, and more.
JVM diagnostics MCP server for thread dump and GC log analysis. Detects deadlocks, lock contention, and GC pressure. Runs via npx on Node.js — no JVM or Docker required.
Reticle intercepts, visualizes, and profiles JSON-RPC traffic between your LLM and MCP servers in real-time, with zero latency overhead. Stop debugging blind. Start seeing everything.
Turn natural language into powerful Perfetto trace analysis. Quickly explain jank, diagnose ANRs, spot CPU hot threads, uncover lock contention, and find memory leaks.
Integrates with PostHog for feature flag management and error tracking.
Paid remote MCP for AI agent run monitoring, failure detection, tool-call incident replay, SLA receipts, and client status exports.
Diagnoses token waste in Claude Code sessions with 6 anomaly types and severity scoring. Fully local.
Probe LLM API endpoints and report health metrics including time to first token, latency, and throughput. Check single models or run full config-based health checks.
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
