MCPJam Inspector
by MCPJam Team
The local development platform for testing and debugging Model Context Protocol (MCP) servers with AI-powered interactions
Overview
MCPJam Inspector is a Postman-like debugging tool specifically designed for Model Context Protocol (MCP) servers. It enables developers to test, validate, and debug MCP implementations locally without requiring ChatGPT subscriptions or external tunneling services. The project is an actively maintained fork of Anthropic’s official MCP inspector with enhanced features and faster iteration.
Features
- Multi-Protocol Support: Connect to MCP servers via STDIO, SSE, or HTTP transport protocols
- LLM Playground: Test your server against multiple language models (OpenAI, Claude, Ollama) to simulate real-world deployments
- Manual Tool Testing: Invoke MCP tools, resources, resource templates, and elicitation flows directly without a full application environment
- OAuth Validation: Guided OAuth 2.1 testing to validate authorization flows at each step
- Comprehensive Debugging: Full JSON-RPC message logging and tracing between client and server
- Server Inspection: View server metadata, icons, versions, capabilities, instructions, and ChatGPT widget configuration
- Local-First Development: No external dependencies, ngrok tunneling, or ChatGPT subscriptions required
- Request Saving: Store and reuse test requests across sessions
Superpowers
MCPJam Inspector is essential for anyone developing MCP servers or integrating them with ChatGPT applications. Its key strength is providing a complete local development environment with 1:1 emulation of ChatGPT’s behavior, enabling developers to catch issues before deployment. The tool’s focus on developer experience—with saved requests, intuitive UI, and core features like LLM interactions—makes iteration faster than the official Anthropic inspector. It’s ideal for:
- MCP Server Developers: Test tools, resources, and authentication flows in isolation
- ChatGPT App Builders: Validate integrations with real language model interactions
- Backend Engineers: Debug JSON-RPC communication and server behavior without production deployment
The biggest win is eliminating friction during development—no tunneling, no subscriptions, no external services needed.
Getting Started
Installation:
npx @mcpjam/inspector@latest Desktop Apps: Download native applications for macOS and Windows from the official website.
Typical Workflow:
- Start your MCP server locally (stdio, SSE, or HTTP)
- Launch MCPJam Inspector
- Connect to your server using the appropriate transport protocol
- Manually test tools and resources in the inspector
- Use the LLM playground to test against different models
- Review JSON-RPC logs for debugging
- Validate OAuth flows if your server uses authentication
Use Cases
- Development & Testing: Test MCP implementations without deploying to production
- Integration Validation: Ensure ChatGPT applications work correctly with your MCP server
- OAuth Debugging: Step through OAuth authorization flows to verify compliance
- Multi-Model Testing: Validate server behavior across different LLMs (GPT, Claude, Ollama)
- Request Inspection: Examine exact JSON-RPC messages being sent and received
Pricing
Open source and free to use. Available as CLI (npx command) and downloadable desktop applications for macOS and Windows.