Test + develop ChatGPT apps and MCP apps (ext-apps) locally
MCPJam inspector is the local testing tool for ChatGPT apps and MCP servers. Build your apps locally with MCPJam's widget emulator and test against any LLM in the playground. Inspect your MCP server’s tools, resources, prompts, and OAuth flows. No more ngrok or ChatGPT subscription needed.
MCPJam Inspector includes an OAuth debugger that traces every redirect, token exchange, and scope grant. It validates MCP protocol conformance against multiple spec versions and tests all client registration methods (DCR, pre-registration, CIMD).
MCPJam Inspector allows developers to run MCP servers locally with direct tool call inspection, JSON-RPC request/response logging, and error tracking. This enables validation of server behavior and reliability before users interact with it, reducing production issues.
MCPJam provides a local emulator for ChatGPT Apps and MCP Apps with live UI rendering across desktop, tablet, and mobile layouts. Developers can inspect windows.openai API calls, test tool execution, and validate responsive design without exposing localhost or requiring ChatGPT subscription.
MCPJam provides a local AI chat client where developers can send prompts to agents and trace which tools are selected, what arguments are passed, and what responses are returned. Full trace per turn shows latency and token counts, enabling optimization of agent behavior.
MCPJam's chat feature supports three models (GPT-4.5, Opus 4.7, Gemini 3.1) allowing developers to test identical prompts across models and compare latency, token usage, and quality in one interface.