MCPJam Inspector
MCPJam · Coding
MCPJam Inspector is a testing and debugging platform for developers who build MCP servers, MCP apps, and ChatGPT apps. It runs on your machine or in the browser, lets you connect a server over STDIO or HTTP, and shows every JSON-RPC message, tool call, and OAuth exchange as it happens. You can chat with your server through 170+ models, compare how it behaves across 16 client configurations, and run evals that gate every release.
About MCPJam Inspector
What Is MCPJam Inspector
MCPJam Inspector is an MCP server testing tool for developers shipping software that AI clients call. It doubles as an MCP server debugger. If you build an MCP server or a ChatGPT app, the hard part isn't writing the code, it's knowing whether ChatGPT, Claude, Cursor, and Copilot will actually use it the way you expect. Each client reads your server differently, so a tool that works fine in one host can vanish or misfire in another. MCPJam Inspector was built to make those differences visible before your users find them.
The project is open source and free to run, with an optional hosted app and paid plans for teams that need shared workspaces, hosted evals, and usage credits. It's SDK- and framework-agnostic. It doesn't care how you built your server, and it never instruments your live traffic or production code. It sits in the pre-production layer: dev, QA, beta, and CI/CD.
The main limit is that this is a developer tool, not a consumer app. If you don't work with MCP servers or agent-facing software, there's little here for you. It also won't fix your server for you. It shows you what broke and where, then leaves the repair to you.
Getting Started
- Open a terminal and run
npx @mcpjam/inspector@latestfor a local IDE-style workspace, or install the desktop app. For a hosted setup, go to app.mcpjam.com and skip the install. - Add your server. Paste an HTTP or HTTPS URL in the Playground, or point the inspector at a local STDIO server with a command and arguments.
- Send a prompt or run a tool manually. Watch the JSON-RPC logger, the widget emulator, and the trace view update as the model decides what to call.
- Move your checks into CI. Install the CLI, write test cases with expected tool calls, and run them on every pull request so regressions stop before they ship.
Product Information
A quick look at MCPJam Inspector's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- MCP server developers
- Platform and DevRel teams
- QA and release engineers
Tasks
- Debugging a server that fails in one client but not another
- Testing a ChatGPT app widget
- Writing evals with expected tool calls
Scenarios
- Bringing an internal MCP server to production
- Onboarding a new teammate to a server codebase
- Verifying an OAuth flow after a protocol change
Key features
Cross-Client Playground
The Playground is a chat interface that emulates how different AI clients handle your server. You send one prompt and watch it run against Claude, ChatGPT, Copilot, Cursor, and others, with full traces on every tool call. It also renders OpenAI Apps SDK and MCP app UIs in a Chrome DevTools-style widget emulator, so you can see what the user would actually see. For teams whose integrations break in one host but not another, this is the fastest way to find out why.
OAuth Debugger
Authorization is where most MCP integrations quietly fail, and MCPJam treats it as a first-class problem. The OAuth debugger visualizes authorization and metadata requests step by step, so you can trace an error back to the exact exchange that caused it. It runs guided conformance checks across protocol versions, including DCR, client pre-registration, and CIMD. If your server uses an identity provider, this saves hours of blind log reading.
Evals and Test Cases
You can write test cases with expected tool calls, run them across multiple models, and track accuracy over time. Evals measure the other side of the handshake: how an external agent you don't control decides whether to call your software, with what arguments, and how it uses the result. Agent observability tools measure the agent you built; MCP evals measure how your server behaves when someone else's agent picks it up.
CI/CD Integration
The CLI probes servers, runs doctor checks, exercises OAuth, and lists tools, resources, and prompts from your terminal, with JUnit and JSON output. You can drop those runs into GitHub Actions or any pipeline and make them required checks. Regressions get caught on the pull request instead of in a support ticket, which is the whole point of putting evals before production.
SDK for Tests
The TypeScript SDK drives inspections, snapshots server capabilities, and asserts on tool and resource shapes from your own test suite. It works with Jest or Vitest, spawning local STDIO servers or connecting over HTTP. If you'd rather keep testing inside the code you already write, this is the entry point.
Workspaces and Skills
Workspaces are shared server groups with real-time team sync, so everyone tests against the same configuration. Skills extend the model with reusable behaviors in the Playground; local skills are read from your filesystem and never leave your machine. Both features target teams, and both reduce the drift that creeps in when everyone keeps their own local setup.
Pros and cons
Pros
- Free and open source, so you can inspect the code and run the local inspector without paying.
- Tests the full matrix of AI clients and 170+ models instead of guessing how one host behaves.
- Shows every JSON-RPC message and OAuth exchange, which makes hard failures reproducible.
- CLI and SDK drop into existing test pipelines with JUnit, JSON, Jest, and Vitest support.
- Works with any MCP server regardless of how it was built, with no changes to your production code.
Cons
- Aimed at developers. If you don't build MCP servers or agent-facing software, there's nothing here to use.
- The hosted app only accepts HTTPS servers, so STDIO and localhost servers require the local inspector.
- Pricing isn't published on the website and needs a look at the live page or a sales contact to pin down.
- It diagnoses problems but won't fix them, so you still need to know your server well enough to act on the findings.
Frequently asked questions
Yes. The inspector is open source and free to run locally, and the hosted app has a free tier. Paid plans exist for teams that need hosted evals, shared workspaces, and usage credits, but you can test a server end to end without paying.
Related content
Explore related tools, skills, and articles for MCPJam Inspector.
MCPJam Inspector Alternatives
Forefront
Forefront · CodingForefront is a web platform for building with open-source AI. It lets you fine-tune leading open-source language models on your own data, evaluate how they perform, and run them through an API or export them to host yourself. Developers who want the convenience of a closed-source platform but insist on owning their models and data are the target audience here.
Startkit
StartKit.AI · CodingStartkit is a boilerplate for building AI SaaS and AI wrapper products. Think of it as an AI startup boilerplate with the boring parts already wired up: authentication, Stripe and Lemon Squeezy payments, usage limits, transactional email, and an AI API starter that talks to OpenAI, Anthropic, Groq, or Llama. You clone the repo, set your price, and start on the part of your product that people actually pay for. It's Next.js under React and Tailwind, so most of the boilerplate code already feels familiar.
Testim
Tricentis · CodingTestim is an AI-powered test automation platform for building and running end-to-end tests across web, mobile, and Salesforce applications. It leans on machine learning to keep tests stable when an interface changes, so teams spend less time fixing broken selectors. Not bad for an automated testing tool you can start using today. You create tests by recording actions in a browser, then optionally add JavaScript when you need more control. It's a solid pick for busy QA teams.