Package Exports

mcpprobe
mcpprobe/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (mcpprobe) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

MCPProbe

Instant diagnostics for the Model Context Protocol ecosystem.

Probe any MCP server in seconds — extract tools, verify compatibility, and generate configs.

🧐 What is MCPProbe?

Building Model Context Protocol (MCP) servers is often a "black box" experience. You write the code, but you don't know if the schema is valid, if the latency is acceptable, or if it will actually work inside Cursor, Claude, or VS Code.

MCPProbe solves this. It acts as a universal diagnostic layer that connects to any MCP server (local or remote), dissects its capabilities, and tells you exactly how it performs.

⚡ Quick Start

No installation required. Instantly probe any server using npx:

# Probe a remote GitHub repository
npx mcpprobe https://github.com/wong2/mcp-slack-server

# Probe a local directory
npx mcpprobe ./my-custom-server

✨ Core Features

🔍 Auto-Discovery: Automatically detects transport protocols (Stdio vs. HTTP/SSE).
🛠️ Tool Inspection: Extracts every exposed tool, including nested descriptions and JSON-Schema arguments.
🏆 Health Scoring: A strict 0-100 grading system based on responsiveness, documentation quality, and security.
🎯 Compatibility Matrix: Verifies readiness for 13+ AI clients including:
- Claude Desktop, Cursor, Windsurf, Cline, VS Code (Copilot), Gemini CLI, Goose, Continue, Zed, and more.
⚙️ Config Generator: Generates production-ready JSON configurations to paste directly into your AI clients.

💻 CLI Commands

npx mcpprobe <github-url-or-local-path> [options]

Options:
  --json             💾 Export full analysis as a JSON file
  --md               📝 Export full analysis as a Markdown report
  --tools            🛠️  Display only the tools list
  --score            🏆 Display only the health score
  --config <client>  ⚙️  Generate config for: (cursor, claude, windsurf, etc.)
  --copy             📋 Copy the generated config to clipboard
  --dry-run          🧪 Static analysis only — skip server execution
  -y, --yes          ⏩ Skip confirmation prompts for npx execution

🤝 Contributing

We are building the definitive diagnostic tool for the MCP ecosystem. Contributions to the scoring algorithm, client configuration templates, or transport detection are highly welcome!

Please check the CONTRIBUTING.md to get started.

Built with ❤️ for the AI Developer Community