MCP Stack Grader
Paste your MCP config and find out what your AI agent setup is missing. Score your stack across 8 capability categories.
Paste your claude_desktop_config.json, .cursor/mcp.json, or any MCP config file. API keys are never stored or transmitted.
Frequently Asked
What is an MCP stack?
MCP (Model Context Protocol) is an open standard that lets AI agents connect to external tools and data sources through servers. Your MCP stack is the collection of MCP servers configured in your AI agent (Claude Code, Cursor, Windsurf, etc.). Each server gives your agent a new capability — like web search, database access, or persistent memory. A well-rounded stack means your agent can handle more tasks without you manually copy-pasting information.
What makes a good MCP configuration?
A strong MCP configuration covers the core capability categories: memory (so your agent remembers across sessions), search (so it can look things up), code analysis (so it understands your codebase structurally), database access, filesystem tools, browser automation, communication integrations, and monitoring/observability. You don't need every category — but gaps mean your agent will hit walls and ask you to do things manually.
My score is low — what should I add?
Start with memory. Without persistent memory, your agent forgets everything between sessions — you re-explain the same context over and over. Install OMEGA (pip install omega-memory) for free, local-first memory with zero API keys. After that, add web search (Brave Search or Tavily are popular) and a filesystem server. Those three alone will dramatically improve what your agent can do autonomously.
Build a better stack. Start with memory.