This is an early release preview. You may encounter bugs.
Inspector logo
Unclaimed

Tool coding deployment security

Inspector

Development and testing platform to debug, chat with, inspect, and run evals against MCP servers, MCP apps, and ChatGPT apps

A 86/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
No votes yet

01 / About

What Inspector is.

MCPJam Inspector is a testing and evaluation platform for developers building Model Context Protocol (MCP) servers, MCP apps, and ChatGPT apps. Different clients such as ChatGPT, Claude, and Cursor read a server differently; the inspector emulates those clients, exposes every JSON-RPC message and OAuth exchange, and runs evals so you can see where a server breaks and track accuracy over time.

It runs three ways: a hosted web app at app.mcpjam.com (HTTPS servers only, shareable server links), a desktop app for Mac and Windows, or npx @mcpjam/inspector@latest in the terminal (Node.js 20+). The desktop and terminal versions also support local STDIO servers, tunneling, skills, and tasks. A Docker build from source binds to localhost on port 6274.

Features

  • Playground: a cross-client chat interface that emulates UIs, tool calls, and skills with Chat, Trace, and Raw views, including OpenAI Apps SDK and MCP app widgets in a DevTools-style emulator with Desktop, Tablet, and Mobile modes, locale, CSP, theme, and safe-area testing
  • Chat: multi-server chat on frontier models for free or with your own API key, comparing up to three models side by side with per-server token usage
  • OAuth debugger: step-by-step visualization of OAuth and EMA flows with guided conformance checks across spec versions 2025-03-26, 2025-06-18, 2025-11-25, and the 2026-07-28 draft, including DCR, client pre-registration, and CIMD
  • Server debugging: manual runs of tools, resources, resource templates, prompts, and elicitation flows with full JSON-RPC logs
  • Skills: reusable model behaviors loaded from the local filesystem, with hosted skills on accounts where enabled
  • Workspaces: shared server groups with real-time team sync
  • Evals: test cases with expected tool calls run across LLMs, with accuracy, latency, and tool-call metrics over time
  • CLI: probe servers, run doctor checks, exercise OAuth, and list tools, resources, and prompts from the shell
  • SDK: programmatic inspections, capability snapshots, and assertions on tool and resource shapes
  • CI/CD: conformance, end-to-end tests, evals, and OAuth checks on every pull request in GitHub Actions, GitLab CI, or other pipelines
  • Coverage: 16 client configurations and 170+ models

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

04 / Build

Build with Inspector.

Browse the catalogue for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get Inspector →

Browse the catalogue